Re: [dns-operations] rate-limiting state
From: =?ISO-8859-1?Q?Colm_MacC=E1rthaigh?= c...@stdlib.net To: Paul Vixie p...@redbarn.org Cc: DNS Operations List dns-operations@lists.dns-oarc.net For example, if the authoritative provider www.example.com were to implement RRL as you describe, then an attacker could spoof traffic purporting to be from Google Public DNS, OpenDNS, Comcast ... etc, and cause www.example.com to be un-resolvable by users of those resolvers. no. it just does not work that way. O.k., so say I spoof 10M UDP queries per second and 10M TCP SYNs per second purporting to be from OpenDNS's IP address. Does RRL a) Let the queries and SYNs go answered. Or b) Rate limit the responses? If it's (a) RRL doesn't prevent the reflection. If it's (b) then you complete a denial of service attack against the OpenDNS users. Which is it? or what's option (c)? I think one option (c) (there might be others) is related to what Paul Vixie meant when he wrote: ] The more common case will be like DNS RRL, where deep knowledge ] of the protocol is necessary for a correctly engineered rate-limiting ] solution applicable to the protocol in http://queue.acm.org/detail.cfm?id=2578510 I've written too many times here and elsewhere that DNS RRL is not a naive firewall rate limit. Simplistic firewall rate limiting against DNS reflections is little better than blocking all ICMP on security grounds. That is why DNS RRL is in the DNS code instead of firewalls. That's also why there are two R's in RRL. There are plenty of words in the documentation, technical reports, and analyses of the various RRL implementations about RRL false positives. There is disagreement about the best values for the parameters that minimize RRL false positives, but we who have the least interest in the topic agree that neither option (a) nor option (b) fit. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] rate-limiting state
From: =?ISO-8859-1?Q?Colm_MacC=E1rthaigh?= c...@stdlib.net I chose a fairly typical number, which is actually below average. Arbor's data on DDOS puts 10M somewhere between the 40th and 50th percentile. I'd be really surprised if OpenDNS's pipes fill up with that kind of small volume. That seems to assume that the infamous Gbit/sec DNS reflection attacks involve one or at most a handful of mirrors. That assumption is wrong. so, third, let's look squarely at large enough UDP flow to activate RRL. 10M requests/sec for www.example.com, type=A. Would that be large enough? 10 Mqps is about 1,000,000 times higher than necessary to trigger DNS RRL. I think 5 or 10 qps is an appropriate DNS response rate limit (although many operators like 50 or even 100). 5, 10, or even 500 qps is a bad limit if your DNS rate limiting is naive firewall counting that pays attention only to source addresses. but I don't think that the numbers work out. If you're getting an attack of 10M PPS, which is very realistic, you'll end up denying service to real users. In most cases (i.e. not OpenDNS, Google, Comcast, etc.), if you're getting 10 Mqps, then your DNS server is denying service to real users regardless of any response rate limiting, because 10M DNS queries/second is perhaps 4 Gbit/sec as well as a healthy CPU load. What is the queryperf number of your DNS system over localhost? (queryperf is a common tool for measuring how many queries your DNS system can answer.) Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] chrome's 10 character QNAMEs to detect NXDOMAIN rewriting
From: Rubens Kuhl rube...@nic.br Yeap, in the source code. Some discussions on those: http://productforums.google.com/forum/#!topic/chrome/dQ92XhrDjfk https://code.google.com/p/chromium/issues/detail?id=47262 http://serverfault.com/questions/235307/unusual-head-requests-to-nonsense-urls-from-chrome http://www.forensicswiki.org/wiki/Google_Chrome#Start-up_DNS_queries If those diagnoses are correct that the probes are for subdomains of the local hosts's domain, then I don't understand grounds to fault Google. The probes should cause no additional traffic to the roots, gTLD servers, or anything outside the user's ISP's network. Whether the local system thinks it is in something bogus like the computer. domain or something reasonable, the Google Chrome probes should hit entries in the ISP's recursive servers's caches. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DNS Caching issue with ATT and Verizon Wireless (other small carriers too)
From: Randy Raitz randy.ra...@readytalk.com I'm writing today after a major mistake by register.com rendered our busine= ss ReadyTalk nearly useless to our customer base. During a routine request = to review our registrar information (removing privacy filters) I'm trying to suppress my kneejerk reaction that people who use spammer shields deserve that kind of trouble and worse. Note that I did not write spam shields, because GoDaddy's original target market of privacy shields were spammers irritated by spam complaints (from spam targets unthinking enough to assume spammers might do something they'd consider good with spam reports) and too stupid or lazy to use one of the many undetectable methods for black-holing complaints to whois contacts. (e.g. throw-away phone numbers, postal addresses, and SMTP addresses that work until the start of an advertising campaign.) Note also that I and my vanity domain have been around for a while, its registration has never used any privacy or proxy services, its registered mailboxes and telephone numbers have always been valid, the current lefthand sides of the mailboxes show I've found it necessary to change them very infrequently, and that some few people claim that I'm hypersensitive to spam. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] summary of recent vulnerabilities in DNS security.
of such very weak security claims are in Haya Shulman's papers.) The reasonable conclusion to this report is Deploy DNSSEC because DNS without DNSSEC is insecure. The paper also contained an explanation of the effect that is unsupported by measurements, tests, or code reading and simply wrong. That explanation is also irrelevant to the point and reasonable conclusion of the paper. The right way to fix the paper is to remove the relevant, unsupported explanation. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] summary of recent vulnerabilities in DNS security.
From: Stephane Bortzmeyer bortzme...@nic.fr Why would there be extra support calls? Wrong keys are no worse than wrong delegations Of course, they are worse. In the vast majority of cases, lame delegations (or other mistakes) do not prevent resolution (as long as one name server works). A wrong key can completely prevent resolution, leading to a loss of service. The DNS is extremely robust, you have to try very hard to break it. With DNSSEC, it's the opposite, you have to be very careful for it to work. Let's agree to somewhat disagree about that. I've found giving one's registrar the wrong IP address or glue a lot worse than a stupid delegation in my own zone files. DNSSEC needs a more effort than plain DNS, but that almost none of extra effort has anything to with registrars. Registrars/registries must sign your DNSSEC RRs, but they must now also sign your other RRs, so that's no extra work for them. Why would registrars get support calls about validation problems? Do they get calls now (that they answer) from DNS resolver operators (other than big resolvers like Comcast) for lame delegations? See above. I cannot visit http://www.онлайн/ while it works from $OTHERISP so it's your fault. I don't understand that. Of course DNSSEC causes more support calls, but the calls are to ISPs and IT groups and not to the registrars trying to sabotage or delay DNSSEC for as long as possible supposedly because of DNSSEC support calls. And again, whether they do suffer more support calls *not*, let them charge extra for adding DNSSEC records! If they can profit from simple DNS for US$10/year, then they could profit with DNSSEC for US$30/year. A rational reason for the registrar DNSSEC sabotage is that the margins on the PKI certs they flog to the punters are at lot more than US$30/year (proof: free certs), and DNSSEC+DANE will eventually kill that cash cow. Yes, no doubt some registrars are too dumb to see that. ... } From: Stephane Bortzmeyer bortzme...@nic.fr } This is not an attack on DNS, but an attack on IP reassembly } technology. } } Frankly, I do not share this way of seeing things. Since the DNS is, } by far, the biggest user of UDP and since TCP is already protected by } PMTUD, I do not think we can say it's not our problem. How dos PMTUD protect TCP? When since perhaps 1995 has PMTUD been seen as protection instead of vulnerability, thanks to goobers with firewalls? Why can't similar attacks using TCP segment assembly be mounted against DNS/TCP? I've heard of more segment assembly attacks than IP fragment assembly attacks, albeit against TCP applications other than DNS. } This might happen even due to malfunctioning network adapter or } other network device, not necessarily an attack. } } A random modification by a malfunctioning device or an errant cosmic } ray has a very small probability of being accepted (UDP checksum, DNS } checks, etc). We are talking here about a deliberate attack, by a } blind attacker. (I thought these latest attacks are not blind, but never mind that.) I've seen more vastly more bit rot undetected by UDP and TCP checksums (esp. due to my own software and firmware bugs and bugs in green hardware) than human attacks. And the number of bad TCP and UDP checksums reported by `netstat -s` on any even slightly busy host should be worrisome. Why does it matter whether the bad bits are natural or man made? Why not turn on DNSSEC, declare victory, and go home? Why continually rehash the falling DNS sky? Aren't there enough other security issues? Some that I've heard about are incomparably worse in consequenes as well as ease of attack (e.g. no hours of 100 Mbit/sec flooding or per-target packet tuning to forge one measly DNS response). Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] summary of recent vulnerabilities in DNS security.
things than fuzz DNS caches with access to the LAN where these tests were done--or most LANS. (By another I'm referring to the mistaken reports that RRL+SLIP=1 is bad because of non-DNSSEC cache corruption under 4 hour 100 Mbit/sec floods.) Instead of looking for yet more obscure ways (e.g. 100 Mbit/sec floods on LANs) in which non-DNSSEC DNS is insecure, why not enable DNSSEC and declare victory? Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] summary of recent vulnerabilities in DNS security.
From: Haya Shulman haya.shul...@gmail.com Please read my first post in this thread, you should find all information there. I see I'm stupid for not seeing that in the first message. I did search for 'http' but somehow didn't see the URL. But why not simply repeat the URL for people like me? Why not the URL of the paper at the beginning instead of a list of papers? https://sites.google.com/site/hayashulman/files/NIC-derandomisation.pdf By searching for DNSSEC with my PDF viewer, I found what I consider too few references to the effectiveness of DNSSEC against the attacks. There is nothing about DNSSEC in the abstract, a list of DNSSEC problems early, and a DNSSEC recommendation in the conclusion that reads to me like a concession to a referee. Others will disagree. After skimming the papers at https://sites.google.com/site/hayashulman/publications since at first I was not sure which one (my fault), I've the impression that Haya Shulman doesn't like: - forwarding to third party resolvers. I agree so strongly that feels like a straw man. I think forwarding to third pary resolvers is an intolerable and unnecessary privacy and security hole. Others disagree. - other mistakes that I think are even worse than forwarders. - DNSSEC Perhaps that will be denied, but I challenge others to read those papers with their litanies of DNSSEC issues and get an impression of DNSSEC other than sow's ear sold as silk. That was right for DNSSEC in the past. Maybe it will be right forever. I hope not, but only years will tell. As far as I can tell from a quick reading, the DNSSEC issues are valid, but are sometimes backward looking, perhaps due to publication delays. For example, default verifying now in server software and verifying by resolvers such as 8.8.8.8 should help the verifying situation. work on DNSSEC improvements and bug fixes before or after your issues? Requiring such answers from me is absolutely out of place, I am probably not aware of the constraints that organisations face in their every day operation of the Internet, and so I never argued which coutermeasures must be deployed and by whom. My goal is to identify vulnerabilities and investigate and recommend countermeasures that can prevent them. Each organisation should decide what solution suites its needs best, based on this and other information that is available to it. That non-answer is absolutely out of place given Haya Shulman's recommendations. It is unacceptable to preassume enough awareness of constraints etc. to tell people 'Do this, that, and the other' but be unwilling to say whether those actions should be done before or after closely related work. This is especially true in this mailing list, because for operators the recommendations are functionally equivalent to do nothing but wait for new DNS software. Port randomization is an extremely thin reed for security, because there are so few port number bits. There are techniques to artificially inflate ports' distribution, and we already described one technique in ESORICS'12 paper. Would that paper be http://link.springer.com/chapter/10.1007/978-3-642-33167-1_16 linked from https://sites.google.com/site/hayashulman/pub ? If so, where or how can I find a free version or a summary of the notion? Getting more than 16 bits of entropy from a 16 bit value sounds interesting. (I trust it's not that literal impossibility.) I've heard of - jumbling domain case, but that suffers from limitations in resolver cache code and it's not part of the UDP port number, - other fiddling with the payload, but they're not the port number, - the ID, but that's not the UDP port number, Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] summary of recent vulnerabilities in DNS security.
I'm puzzled by the explanation of Socket Overloading in https://sites.google.com/site/hayashulman/files/NIC-derandomisation.pdf I understand it to say that Linux on a 3 GHz CPU receiving 25,000 packets/second (500 bytes @ 100 Mbit/sec) spends so much time in interrupt code that low level packet buffers overflow. That puzzles me for reasons that might be summarized by considering my claim of 20 years ago that ttcp ran at wirespeed over FDDI with only 40-60% of a 100 MHz CPU. https://groups.google.com/forum/#!topic/comp.sys.sgi.hardware/S0ZFRpGMPWA https://www.google.com/search?q=ttcp Those tests used a 4 KByte MTU and so about 3K pps instead of 25K pps. The FDDI firmware and driver avoided all interrupts when running at speed, but I think even cheap modern PCI Ethernet cards have interrupt bursting. Reasonable network hardware interrupts the host only when the input queue goes from empty to not empty or the output queue goes below perhaps half full full, and then only interupts after a delay equal to perhaps half a minimum sized packet on the medium. I wouldn't expect cheap PCI cards to be that reasonable, or have hacks such as ring buffer with prime number lengths to avoid other interrupts. Still, ... IRIX did what I called page flipping and what most call zero copy I/O for user/kernel-space copying, but modern CPUs are or can be screaming monsters while copying bytes which should reduce that advantage. It would be irrelevant for packets dropped in the driver, but not if the bottleneck is in user space such as overloaded DNS server. That old ttcp number was for TCP instead of UDP, which would be an advantage for modern Linux. So I would have guessed, without having looked at Linux network code for many years, that even Linux should be using less than 20% of a 3 GHz CPU doing not only interrupts but all of UDP/IP. 100MHz/3GHz * 60% * 25000 pps /3000 pps = 17% Could the packet losses have been due to the system trying to send lots of ICMP Port-Unreachables? I have the confused impression that Socket Overloading can involve flooding unrelated ports. How was it confirmed that kernel interrupt handling was the cause of the packet losses instead of the application (DNS server) getting swamped and forcing the kernel to drop packets instead of putting them into the application socket buffer? Were giant application socket buffers tried, perhaps with the Linux SO_RCVBUFFORCE? (probably a 30 second change for BIND) 25K qps is not a big queryperf number, which is another reason why I don't understand how only 25K UDP qps could swamp a Linux kernel. Just now the loopback smoke-test for RPZ for BIND 9.9.4 with the rpz2 patch reported 24940 qps without RPZ on a 2-core 2.4 GHz CPU running FreeBSD 9.0. What about the claims of Gbit/sec transfer speeds with Linux? https://www.google.com/search?q=linux+gigabit+ethernet+speed I'm not questioning the reported measurements; they are what they are. However, if they were due to application overload instead of interrupt processing, then there might be defenses such as giant socket buffers. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] summary of recent vulnerabilities in DNS security.
elevate a convenient issue like raw sockets above all others. To the extent that the punters paid attention Steve Gibson's campaign against Windows XP because of raw sockets, he harmed them and the Internet in general, because XP was less insecure than previous versions of Windows. } Furthermore, if port randomisation is not an issue why standardise } [RFC6056]? Why set up DNS checkers? If current port randomisation } algorithms are vulnerable - why not fix? That's a straw man. No one has said that ports should not be random. It was an unfortunate oversight that RFC 1948 did not recommend random emphemeral ports. If RFC 1948 had mentioned port numbers it would have concluded: Good sequence numbers AND PORT NUBMERS are not a replacement for cryptographic authentication. At best, they're a palliative measure. (capitalized words added to actual RFC 1948 text). Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] summary of recent vulnerabilities in DNS security.
From: =?ISO-8859-1?Q?Colm_MacC=E1rthaigh?= c...@stdlib.net Economics also include costs. The operational cost of deploying DNSSEC validation on resolvers remains high - there are still frequent key rotation and signing errors that cause various DNS subtrees to be unresolvable. On what do you base your claims about the fatal costs of DNSSEC validation? I claim relevant knowledge and experience, not just from code I wrote a few years ago to reduce the costs of DNSSEC on very large resolvers, but from signing my own domains and enabling validation on all of the resolvers that I control. My domains and resolvers are insignificant, but I hope I would have noticed any fatal costs. Are you aware that Comcast's resolvers have been validating for some time? I think Google is also validating based on a Webmaster, your web page is not available to your spider messages after a configuration error in my signing machinery, but am not sure. Does that conflict with your claims about the fatal costs of validating? Yes, I've noticed that Google is still not signing. Maybe the continuing hijackings of their ccTLD domains will move them. If an attacker can cause the domain to be unresolvable, that seems like a weakness. True, but the right question is not Does DNSSEC add vulnerabilities? but Overall, is DNS more or less secure with DNSSEC? or Among all of the things I can do, what will improve the security of my users and the Internet in general? Defenders who care about the security of their systems and the Internet in general don't pick and choose among weaknesses based only on what is easiest, what can be punted to others, or what contributes to their reputations. They don't do as Steve Gibson did and harp on the bogus catastrophy of Windows XP raw sockets to enhance his reputation and sell his services. Kaminsky wasn't the discoverer of the Kaminsky's bug either, it was long known, yet here you credit him. Not that I mean to deny credit to Kaminsky, he did a good job of publicising the vulnerability. Just as Haya has done here. I suspect Kaminsky got the credit because he had been contributing to the field for years. But who cares who got there first? Every request I see for credit is recorded in my private accounting as a debit against the credibility of the person demanding credit, because credit demands suggest interests which suggest biases and so inaccuracy. Yes, I've heard of Kaminsky's business interests, and so I don't take his announcements at face value. You should also discount my credibility based on my pecunary or other interests. Where you can't determine my interests, act on your best guess. Back before Kaminsky made the need for port-randominsation undeniable with an actual working PoC, this sounds like the ISC/Bind response to port randomisation attacks. Other implementors and operators made a better judgement avoided the problem entirely, taking the cautious path. 5 years later, are you really saying we should ignore another attack vector? Who besides you and Haya Shulman has said anything about not randomizing ports? What port randomization improvements do you think are needed in current releases of any major DNS implementation? Where port randomization problems exist such as in junk CPE that won't get fixed before I retire, what contributes most to solutions, selling $29.95/#24.95/£19.95 academic papers or turning on DNSSEC? The issue for me is one of relative priorities. Among all Internet security issues that I might touch, which should get my attention and effort? By remaining silent about emphasising port randomization over DNSSEC (or using distant instead of nearby validating resolvers) would I help or harm? The impact even with DNSSEC fully enabled seems concerning enough to warrant attention. Let's agree that ports ought to be as random as TCP ISNs, improve port randomness where each of us can, and stop implying that anyone thinks or says otherwise. Let's also stop the DNSSEC is a problem stuff. Finally let's consider how you are helping. Is there anything you can do to improve port randomization? If you are committer in any open or proprietary source trees, will you make any needed port randomization fixes? Have you deployed DNSSEC? What about BCP 38, since cache poisoning is likely to depend on BCP 38 violations? Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] summary of recent vulnerabilities in DNS security.
From: Warren Kumari war...@kumari.net I suspect they're more interested in getting registry lock in place rather than DNSSEC. Most of the attacks against Google have involved changing the name servers completely .. Through social engineering and sometimes through directed attacks, yes. Sadly yes. I trust we all agree that cache attacks with non-random ports, fragmentation, or padding are irrelevant except perhaps indirectly through the general (lack of) value of DNSSEC that I claim better prevents cache attacks than random ports. Wouldn't DNSSEC have not made things worse and possibly made them better by: - making the social engineering more difficult by forcing the bad guys to change key as well as NS RRs - possibly making the bogus records fail to validate for a while at the start of the attack, thanks what might look like an unplanned KSK change. - possibly making the bogus records fail to validate sooner and so get ignored sooner after the registrar records are restored, again thanks to what might look like an unplanned KSK change. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] summary of recent vulnerabilities in DNS security.
factors. Port randomisation algorithms that were proposed have weaknesses, but proper randomisation should solve these problems. I doubt the implication in that sentense, because I give you more credit than to think that any of the recommendations in your 19 Oct 2013 mail could and would be deployed overnight or easier than DNSSEC by any except a very few individuals. The only recommendation that might be done quickly is forcing TCP to proxies. That could be done with a firewall ACL against UDP, but no one should do that. The others require changes to source. For example, several of your recommedations are to randomise ports. Any code that doesn't already do that will continue not doing it at least until its next release. Fixing your vulnerabilities would be technically easy for many people, including those whose resumes are like mine. However, what is easy for professional kernel and DNS server hacks is irrelevant. I was under pressure to catch a flight when I responded and forgot DNSSEC; it is as dear to me as it is to you :-) I'm sorry, but I think the mention of DNSSEC in your paper exists only because others forced it. I'm forced to that belief by various things including your refusal admit the obvious about relative priorities and by statements like that sentence above that suggests that fixing port randomization could be easier than deploying DNSSEC in any except quite exceptional cases. You directed Paul Vixie to a copy of your paper on your web site, but I didn't see a URL. Is http://www.ec-spride.de/ your web site and if so, is your paper there? I can't find any other likely URLs. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] summary of recent vulnerabilities in DNS security.
From: David Conrad d...@virtualized.org Should the people working on DNS implementations prioritize making their DNSSEC code more robust and easier to use above or below addressing your issues? I'd say below. Resolver operators (hopefully) want to protect their caches. DNSSEC = will do that, but only if people are signing their zones. There are lots = of external parties (e.g., registries, registrars, software developers, = resolver operators, etc) to get DNSSEC deployed and there remains very = little incentive for anyone to sign their zones, regardless of how = robust and easy it might be made. The alternative would be to disregard current and future cache poisoning = attacks. Pragmatically speaking, I personally think it highly = questionable to ignore cache poisoning vulnerabilities because something = which isn't yet deployed to 10% of the Internet will fix it. This would be a bit like saying don't deploy RRL because BCP38 is the = correct answer to the problem. On the contrary, anyone who spends even one minute on RRL that could be spent on BCP 38 should be...well, I can't say shot because I oppose capital punishment. RRL should be considered only after everything possible has been done for BCP 38. Similarly, only after there is nothing that you can do improve your DNSSEC implementation should you consider improving your port randomization. I agree that port randomization should come before a lot of other things, although that's not saying much because the major DNS implementations are filled with things I would have vetoed if I'd been king. I think their work showing the weaknesses of port randomization in theory and practice is important, because it shows that no security should depend on adversaries being unable to inject packets into UDP or TCP streams because ports are secret. I strongly disagree with Haya Shulman's words to Paul Vixie that seemed to say that their work might fix other applications and protocols. I think their work shows that port randomization is like RRL, a lame kludge of a mess that is better than nothing but not even a distant second choice to actually fixing the problem. I say only consider improving port randomization, because nothing should be added to anything or even changed without clear and significant benefits, especially in security related areas. You've been around long enough to remember many added nice features caused big security problems. Vernon Schryverv...@rhyolite.com P.S. I'm licensed by http://ss.vix.su/~vixie/isc-tn-2012-1.txt and http://ss.vix.su/~vjs/rrlrpz.html to criticize RRL. P.P.S. I've often heard Paul say much the same thing about RRL being a bad idea except compared the alternative of ignoring the consequences of everyone else's failure to deploy BCP 38. ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] summary of recent vulnerabilities in DNS security.
From: Haya Shulman haya.shul...@gmail.com We (me and my phd advisor Prof Amir Herzberg) recently found a number of new DNS vulnerabilities, which apply to patched and standard DNS resolvers, ... Recommendations: ... The complete absense of any mention of DNSSEC among those recommendations (or elsewhere) reads like an implicit claim that DNSSEC would not help. Even if that claim was not intended, would it be accurate? Would DNSSEC make any of recommendations less necessary or perhaps even moot? If DNSSEC by itself would be effective against cache poisoning, then isn't it among the recommendations, especially for Resolver-behind-Upstream? Why aren't efforts to protect port randomization, hide hidden servers and so forth like trying to make it safe to use .rhosts and /etc/hosts.equiv files by filtering ICMP dedirects and IP source routing, and strengthening TCP initial sequence numbers? It's not that filtering ICMP redirects, etc. are wrong, but I think today those things are used for availability instead of data integrity (or authentication and authorization), and small leaks are not always and everywhere seen as catastrophes. In fact, haven't ICMP redirects been reborn as fundamental parts of IPv6? Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] summary of recent vulnerabilities in DNS security.
From: Haya Shulman haya.shul...@gmail.com IMHO, DNSSEC is simply the natural defense against the attacks, which is why I did not explicitly mention it, but I definitely had it in mind :-) In that case, on what should an organization spend time or money first, on DNSSEC or the recommendations in the mail message? Would it be better if each of the recommendations in the mail message started with something like this? Deploy DNSSEC, and consider the follow to help protect cached data not yet protected with DNSSEC. Regarding the proxy-behind-upstream: to prevent the attacks DNSSEC has to be deployed(and validated) on the proxy. Currently it seems that there are proxies that signal support of DNSSEC (via the DO bit), but do not validate responses, and validation is typically performed by the upstream forwarder. That sounds like a more significant bug than port obscurity or randomization. If it is a bug, which should be addressed first in that software or those installations, this DNSSEC bug or the recommendations in the mail message? It it is a significant DNSSEC bug, it would be good if a future version of the mail message mentioned it. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Should medium-sized companies run their own recursive resolver?
sizes, etc.. and you're in the right range for 4-5+ devices per home at least. So 500 unique stub resolvers is a large population, but at the same time, it's not enough to justify spending $1k for a server *2 + UPS. That cost on my friends WISP would wipe some or all profits. Perhaps he should charge more, but he doesn't have the knowledge or skill to operate things, but does know it might help to have a faster box on-net That's yet another nonsensical and offensive (because it insults the reader's intelligence) red herring. That organization doubtless has a firewall, often part of a router. That firewall could easily run a recursive resolver for those 500 stubs, because DNS requires only a fraction of the resources needed for firewalling or routing. The configuration of that recursive resolver could be entirely automatic, because the firewall has all of the parameters needed to configure a closed verifying recursive resolver. Yes, your Open Resolver project might have found that many borken open DNS forwarders in CPE routers and bridges answer WAN requests. The fault for that lies with your big ISPs with those big, creative DNS resolvers, because they chose the incompetent, below cost CPE vendors, and shipped and supposedly continue to support and maintain that junk CPE. A dispassionate observer might see that as a compelling argument against using more products from your big ISPs, such as their resolvers. I am offended on behalf of those hypothetical IT professionals by your persistent infantilizing them. Attitudes like yours in ISPs are why there there is so little BCP38 compliance and so many open resolvers. Ha! BCP-38 is actually *hard*. Customers don't know their IP ranges or domains they own/operate. That's yet another irritating and offensive red herring. As you know, BCP 38 is done in the vicinity of routers. Customers who don't know their IP address blocks aren't running their own routers and aren't responsible for BCP 38. Even when customers are configuring their own BGP, the primary responsibility for BCP 38 is upstream at the ISP. If ISPs would refuse to route packets to customers that can't comply with BCP38 or that run unnecessary open resolvers or open resolvers unprotected by rate limiting, then a lot of problems would go away. I encourage you to start an ISP and report back your results :) I can either forward packets or filter them in many cases. The damaging packets, even when there are large attacks are still such a small percentage of the overall traffic that I can't sacrifice them for the sake of others. You've seen other providers back away from BCP-38 over the years. I continue to strive in this direction and I resent the fact that you think we're all not trying to do it. So an ISP expectations of riches justifies taxing the rest of the Internet to deal with the ISPs abusive customers? For years, many ISPs and their employess said the same things about their spamming customers. Those that didn't respond to spam reports with Just hit delete and stop bothering me sonny demanded that I dontate my time to them (ISPs) by making spam reports. Both tactics were and were intended as substitutes for dealing with spamming customers. ISPs whined about losing customers should they outlaw outgoing spam. Some such as MCI/UUNET spun lies about the technical impossibility of counting outgoing TCP SYNs to port 25 as reasons why MCI/UUNET could not find spam friendly resellers. Today the spam problem is largely solved, partly by better filters, but also by ISPs discovering that ignoring spam is unprofitable. If you think you can build a better router that can do it right, please do. The market needs it, but that's OT for this list. That's yet another irritating and offensive red herring, because I said nothing about building better routers and because better routers are not needed for BCP 38. What's needed is equivalent of the spam solution, de-peering ISPs that can't be bothered. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Should medium-sized companies run their own recursive resolver?
From: Carlos M. Martinez carlosm3...@gmail.com Also, customer CPE equipment is poor and ... Agreed. CPEs cannot be trusted. That fact is a poor argument for trusting the recursive resolvers of the organizations responsible for that worse than junk CPE. Most of that worse than trash CPE is specified, tested, provisioned, and maintained by the same outfits. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Should medium-sized companies run their own recursive resolver?
From: Jared Mauch ja...@puck.nether.net phones, and other devices behind a NAT router owned by and remotely maintained by Comcast. Instead the question concerned a business with 2 IT professionals. Relying on distant DNS servers is negligent and grossly incompetent for a professionally run network. As with many things we will have to disagree. Not everyone has the same skill set as those on this list, and that curve goes down rather quickly. I can't help noticing that Jared Mauch noticed and disagreed with my conclusion about relying on distant DNS servers but overlooked or ignored the security reasons compelling the conclusion. He evidently also overlooked the contradiction or irony in his previous note: ] Everyone else should just use either their ISP (with NXDOMAIN ] rewriting turned off) ... ] Folks like Comcast have large validating resolvers. Their customers ] should use them. despite https://www.google.com/search?q=COMCAST+dns+hijacking If you check the pages found by that URL, you'll see - older reports that Comcast was phasing out DNS hijacking - more recent reports of redirection or hijacking of 58/UDP packets--not just falsified results from those big Comcast DNS servers but packet hijacking - far more complication, confusion, and mystification than is realistic to expect a two person IT department to resolve. It's clear that a simple, securite business DNS configuration does *not* involve a consumer grade ISP. (I don't mean to criticise any particular consumer grade ISP. They are all similar. I'm not even sure that DNS result or packet hijacking is a bad thing for consumer households.) However, not just tolerating but encouraging people without basic network and computer competence run Internet businesses is like aviation before the FAA. In the first years enthusiasts bought, built, or borrowed airplanes and went into the barnstorming or airmail businesses. Then the air industry got government licenses and regulations. From Kitty Hawk to the 1926 Air Commerce Act licensing pilots was 23 years. http://www.faa.gov/about/history/brief_history/ Whether you mark the start of public interest in the Internet with the 1972 CACM articles about the ARPANET (my DOC lab employer read those papers, got an appropriation, and linked our computers soon after), CSNET co in the early 1980s when many commercial outfits with got Internet connections, or a date between, it is more than 23 years later. I don't like the idea of government Internet licenses, but a two person IT shop using distant DNS servers, not to mention a consumer grade ISP, is as culpable as buying an old potato washer to clean your cantaloupe crop for market. I'm uncomfortable with the criminal charges against the Jensen brothers, but if that's what it takes to get people learn enough and do it right ... https://www.google.com/search?q=Jensen+cantaloupe Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Should medium-sized companies run their own recursive resolver?
From: Bob Harold rharo...@umich.edu I think the problem with a DNS appliance is that it becomes an open DNS resolver, unless it is configured to know the subnet(s) used internally, and updated every time that changes. I don't think the firewall could reasonably be asked to block only recursive DNS traffic, although perhaps it could block all inbound DNS requests, except to an internal authoritative DNS if you had one. I cannot think of any other simple workaround. Users are likely to find some way to turn off the recursion limiting anyway, like setting the internal subnet to 0.0.0.0/0, which solves their problem of updating it when subnets change, but leaves it open to the world. There is a trivial and easy way to keep a recursive DNS server intended for an organization with a 2 person IT departement from being open to the entire Internet. Set the IP TTL on responses both TCP and UDP to a small number such as 3 or 5. There are business reasons to keep a small DNS appliance intended for a small business with a 2 person IT department from being used by a big outfit. You might limit the number of DNS responses per second, hour, or day, but it might be better instead or also to limit the number of client IP address. It would be trivial and easy for a DNS appliance to require ACLs permitting no more than X IPv4 addresses and Y IPv6 /64's. Ship it configured with 10.0.0.0/8 and have it refuse to accept non-RFC 1918 ACLs with too big a total. A little monitoring of requests from unexpected IP addresses and some GUI sugar would make it easier for users to maintain their ACLs than what I've seen in the DNS, AD, WINS, etc. settings of a Windows box. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Should medium-sized companies run their own recursive resolver?
From: Jared Mauch ja...@puck.nether.net ... Mercedes... Have you ever driven one? They are mighty nice :) Back in the 90's I would agree everyone should run a DNS server as the network wasn't as robust as it is today. On the contrary, in the relevant sense, the network today is less robust than it has ever been. You don't want a commodity luxury sedan while driving across Syria, Iraq, Afghanistan, or the Gobi Desert despite the fact that many roads in Europe and N.America are more robust than they've ever been. Where roads are bad or non-existent or where there are significantly security hazards, you need something with more armor, ground clearance, spare fuel, water, emergency supplies, or even guns than are economical or safest elsewhere. Some folks may need local elements (e.g.: MS DNS/AD, but these should not be exposed to the internet... Everyone else should just use either their ISP (with NXDOMAIN rewriting turned off) or someone like OpenDNS that can help enforce some security policies and practices with a few knobs being turned at most. Folks like Comcast have large validating resolvers. Their customers should use them. Folks here are surely going to do the right thing the majority of the time. The vast majority of others are going to set things up once and it *will* be left to rot. This isn't intentional, but it naturally happens. The question had nothing to do about J. Sixpack with 37 televisions, phones, and other devices behind a NAT router owned by and remotely maintained by Comcast. Instead the question concerned a business with 2 IT professionals. Relying on distant DNS servers is negligent and grossly incompetent for a professionally run network. When the DNS servers in question are to known lie, it should be as much a crime as failing to wash your cantaloupes in Clorox. https://www.google.com/search?q=COMCAST+dns+hijacking https://www.google.com/search?q=jensen+farms+criminal The same applies when there are Great or small firewalls between the DNS client and distant validating recursive resolvers. Even Joe and Joan Sixpack should, if they can, think carefully about relying on distant DNS servers. If you wouldn't give your ISP your bank passwords, then you shouldn't rely on your ISP to validate your RRs. Those who control your RRs can get your passwords, albeit with varying effort. Should Joe and Joan rely on government approved DNS servers while they are in China, Iran, or Syria? Never mind that if the U.S. NSA, FBI, CIA, etc. are competent, they've used DNS creatively such as to install software on the computers of their targets or deploy MX RRs to monitor email. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Microsoft
From: Jim Popovitch jim...@gmail.com Ha! I removed one ~6 months ago and since then I've been 550 rejecting the reports... yet they still come in. Oh wow. It was more than 9 months ago (_dmarc.spammers.dontlike.us was removed on 15-Jan-2013). I saw something similar from Microsoft while playing with DMARC. Microsoft never forgot _dmarc records I simply deleted. However, publishing records with reporting or checking explicitly turned off were eventually effective. I think it might have taken a week for the reports to stop. This might sound like a bug or problem with DMARC at Microsoft, but it might be a feature implied by the same design requirements the cause DMARC to apply SPF DNS records more broadly than RFC 4408 allows. For example contributions to this mailing list from domains using DMARC+SPF with rejection will not be seen at free Google or Microsoft mailboxes, because the SMTP envelope Mail_From value will not be in even relaxed alignment with the From: field in the forwarded contributions. It might even be a feature instead of a bug that according to my tests, if you configure Microsoft mailbox to foward to Gmail mailbox or vice versa, the forwarded mail will not be delivered. Contrary to what one might guess from https://tools.ietf.org/html/draft-kucherawy-dmarc-base-01 https://en.wikipedia.org/wiki/DMARC and http://www.dmarc.org/overview.html DMARC seems intended to improve communications between large scale mailbox providers such as Microsoft and Google and bulk mail senders. DMARC tells bulk mail advertisers such as American Greetings and Linkedin about inbox placement. It tells bulk mail senders might prefer their bulk mail not be forwared such as Fidelity Investments and JPMorganChase. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Implementation of negative trust anchors?
From: David Conrad d...@virtualized.org Exactly so. However pragmatically speaking if someone (say NASA = perhaps?) screws up signing their zone, it isn't the = zone-signing-screwer-upper that gets the phone calls, it is the eyeball = networks that are doing the validation. Without NTA, the eyeball = network operators have a choice, eat the cost of those calls or turn off = validation _for ALL signed zones until the zone-signing-screwer-upper = fixes their problem_. I gather you believe eating the cost is the right answer. =20 YES! Eyeball networks are paid by their customers to act as pre-front-line support for bad DNS delegations, broken HTTP servers, and all other content provider problems. Saying otherwise for any of the services sold by eyeball networks is another step down the slope toward content providers paying eyeball networks for eyeballs and the conversion of the Internet into what it was in about 1965 when it was owned by Ma Bell and the three television networks. Of course, it wasn't called the Internet, but it was the contemporary equivalent. I was around for the Carterphone decision and the incredible freedom to connect computers that followed soon after (in about 15 years--remember DAAs?). I was also around to see the ARPANET use 56kbps leased lines that were not only incredibly slow but required incredible massaging of Ma Bell bureaucrats who required you to admit who was in really charge of your business. (I was at TIP-25 at DOCB) } From: David Conrad d...@virtualized.org } Vernon, } If the only solution to someone else screwing up signing is to turn off = } validation for all zones and the likelihood of someone screwing up = } signing scales with the number of folks signing, why bother ever turning = } validation on? Eyeball networks would be best served by turning off DNSSEC. Comcast not withstanding, DNSSEC does nothing to help their bottom lines. Let's be honest and admit that talk about NTA today and tommorow (as opposed to last year) is really a statement of regret about DNSSEC and a demand that DNSSEC just go away. If you honestly believe in DNSSEC's promise of letting me sign my zones, then you must also let me mess them up. Essentially none who will use NTA will have any inkling whether bad signatures on my zones reflect my incompetence or actions of my (and or their) enemies. Many of us here now can and are happy to make good guesses about whether a DNSSEC failure is due to zone operator error or enemy action, but that won't be true of most future NTA users, including big outfits. I read the thinness of http://dns.comcast.net/ as saying that Comcast, that major NTA supporter, has not only given up trying to diagnose other people's DNSSEC problems but quietly shelved NTA. } On the contrary, NTA is a new tool for deliberately introducing new } faults in the data you give your DNS clients. } True. This is why I suspect corporate types will have hesitancy to use = } NTAs and wish to remove them as soon as possible. On the contrary, given minimal cover such as an RFC, corporate types at eyeball networks will mandate add-only NTA lists that only grow and never lose entries. They'll say politically correct things about DNSSEC but use NTA to minimize support costs and maximize profits from activies that are incompatible with DNSSEC such as typosquatting. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Implementation of negative trust anchors?
From: Evan Hunt e...@isc.org On the contrary, given minimal cover such as an RFC, corporate types at eyeball networks will mandate add-only NTA lists that only grow and never lose entries. Obviously that's possible, but IIRC the draft requires that NTA entries have limited (and short) lifetimes. HAH! If RFCs were Law, then the DNSSEC RFCs would have long since answered any question about NTA as ABSOLUTE NEVER! In the real world, RFCs are no more or less than hints on what to do to minimize complaints and sanctified excuses for doing what you want to do anyway. If we decide to implement this in BIND (it's on our roadmap, but with a question mark), I expect the NTA lifetime will default to an hour and be capped at a day. NTAs would be inserted via the control channel (rndc) rather than a configuration file change, and wouldn't persist across system restarts. An operator could write a script to continually insert the same NTA's over and over again forever, but it would be easier to allow them to lapse as intended. I agree that's not nearly as evil as NTAs in a configuration file, or a cron script that runs every 30 minutes and does a few 100K `rndc nta` commands to fix that problem that someone reported year before last in the .gov signatures, and protect the advertising revenue from those typosquatted domains. I was against NTAs when they were first proposed; I've come around. Disabling validation because of signing failures is the wrong thing to do, but people are going to do the wrong thing whether I like it or not, and if we must choose between evils, I prefer rndc validation off nasa.gov to rndc validation off. On the contrary, in the real world this year, the people using `rndc nta` will decide after the 42th time in 48 hours of renewing the protection for the .gov problem (not counting the 6 renewals that should have been done between 01:00 and 03:00 when the people empowered to use `rndc nta` were asleep) to either `echo rndc nta nasa nta-cron-script` or `rndc validation off`. Next year, those empowered peole will be tired of diagnosing DNSSEC problems and arguing with their bosses about value of DNSSEC. They'll give second-line support a button to push that does `echo rndc nta $1 nta-cron-script`. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Implementation of negative trust anchors?
From: Evan Hunt e...@isc.org it or not, and if we must choose between evils, I prefer rndc validation off nasa.gov to rndc validation off. ... } A document that advised limits on the use of NTAs -- for example, the } recommendation in Jason's draft that they not persist for more than } a day -- would be okay by me. On second thought, Consider the situations of resolver operators confronted with a situation where you might use `rndc nta`. Almost all of them will (and even now most) lack the expertise, time, inclination to figure out which domain to hit with `rnd nta sub.dom.example.com`. They'll only know (or hope) that the irate phone calls from principals about broken lesson plans are related to DNSSEC problems. They would be better served by `rndc validation off X hours` with a limit on the X hours of 24 than any sort of NTA hook. If you don't let them to use `rndc validation off X hours`, most will use `rndc nta gov` because their users will be shouting about governement web site problems and they won't have the time, inclination, or permission to discover that it's only the apod.nasa.gov. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Implementation of negative trust anchors?
From: wbr...@e1b.org and protect the advertising revenue from those typosquatted domains. Why would a typosquatted domain fail DNSSEC? When DNSSEC is universal and easy to do, it will be signed from the TLD on down, just like every other domain. I wasn't talking about the typosquatters who give the registrar and registry their cuts. I meant the typosquatters who (at first) replace only NXDOMAINs (and later decide that what's good for the legitimate typosquatters is good for them). I assume we all remember the NXDOMAIN typeosquatting kerfuffles. Are any (potential) eyeball networks (in hotels for example) squatting on NXDOMAINs today? Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Implementation of negative trust anchors?
From: David Conrad d...@virtualized.org They would be better served by `rndc validation off X hours` with=20 a limit on the X hours of 24 than any sort of NTA hook. So, because one zone messes up signing, instead of opening up that one = zone to spoofing attack you think it is better the resolver operator = opens up all zones to spoofing attack? This seems wrong to me. It's wrong only if you accept the false choice between validation off and a targeted NTA. We're talking about *resolver* server operators, not authority operators or IETF participants. Big resolver server operators not selling resolution will not bother figuring things out. They'll ignore complaints, send users chasing whois phone numbers, or turn off DNSSEC. They don't have time or permission to diagnose other people's DNSSEC problems enough to use NTA correctly. See the Comcast web page for proof of that. The resolver servers selling resolutions will use NTA correctly, but they already have NTA and don't care about opinions from peanut galleries including the IETF. The majority of resolver server operators will not use NTA more than a half a dozen times. Then they'll treat DNSSEC errors like bad delegations or use one form or another of validation off including NTA as close to the root as they can go. The best bet to keep them from a static validation off is an automatically sunsetting form. I'd suggest that in the BCP/RFC/whatever, in addition to recommending = that NTAs be time capped and not written to permanent storage, it should = also recommend NTAs be written as specifically as possible. Yes, that transient NTA a good idea I'd not heard/noticed/understood until today, but it does not redeem NTA. I can't believe you're seriously suggesting that words in any IETF document telling people to use narrow NTAs would have any effect on resolver operators. Practically no one who might use any NTA hook will understand or (be allowed to) care enough to figure out to hit cnn.co.uk instead of cnn.com. Of necessity they'll just keep hitting the NTA button with semi-random domains until the calls stop. The wise ones will go straight as high as they can, functionally to validation off. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Implementation of negative trust anchors?
From: wbr...@e1b.org Running the DNS for 100+ school districts and 400,000+ devices, I really, REALLY don't want to be the one saying Sorry, you can't use the site called for in your lesson plan today because they messed up the DNSSEC records. Management's response would be Just make it work! Without a per domain NTA, the only option would be to turn off DNSSEC, returning to square one. You don't do crazy things like poke around to get an old copy of their zone and publish a pirate copy when they mess up something else. You say something like They messed up. In this case, you could and should say something like: Our network security defenses are telling us that there is something. wrong there. Instead of lesson plans, you might be getting child porn if you visit their pages today. Our browsers give us the option to trust invalid TLS certificates, some even storing it indefinitely. Is an NTA much different? Yes, because TLS differs because public PKI certs are merely a charade of pretend security intended to fool the rubes and harvest money from those cannot for various good and bad reasons refuse to pay the commerical PKI cert vendors. (Yes, some commercial PKI certs are free, which says all that needs to be said to anyone with 0.1% of a clue about the security of every commercial PKI cert.) A valid commercial PKI cert tells you *NOTHING* about the web data it purports to guarantee except that some was willing to pay time, effort, and perhaps some money to appear trustworthy. Perhaps in the real world, no evil nasty hackers are going to replace your staff's educational pages with nastiness with either bogus certs or corrupt DNS, but things are definitely otherwise elsewhere. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Implementation of negative trust anchors?
From: Suzanne Woolf wo...@isc.org I don't like it either, but it limits the damage done by a DNSSEC = failure to status quo ante rather than something worse. That is mistaken. You get the status quo ante by simply turning off validation. Turn off validation is the only sane response this year to phone calls reporting the breakage of a major domain. Even if you have NTA, from now on you'll do as Comcast evidently is now doing and decline to pay the current and future costs of adding minor domains to your NTA list. You'll just tell your users Stuff Happens and perhaps help them use `whois` to find someone else to bother. Last year differed. I trust (wish?) we all learned the excessive costs of organization-wide white/blacklists from the last 15 years of the spam wars. madness test: would we have bothered with DNSSEC at all, back in the = day, if NTA had been known as a definite requirement? I realize this is something of a rhetorical question, but I'll bite: if = it were framed as a way of promoting incremental, fault-tolerant = deployment and mitigating the cost shifting of I screw up and your = phone rings, some of us might well have been happy to include it.=20 On the contrary, NTA is a new tool for deliberately introducing new faults in the data you give your DNS clients. It is a tool for lying to your DNS clients with data that you swear is valid and signed but that you know is at best unsigned and quite possibly invalid or worse. If I didn't know that the inevitable user response to security problems, I'd favor NTA as a way to get validation move where must be eventually, at least as close as their nearest router. After a few kerfuffles in which it is discovered that telephants have been ordered by government or corporate bosses to use NTA to obscure the hijacking of domain names on grounds of copyright violation, terrorism, publication of national defense secrets, or failure by content providers to agree to telephant tariffs, one might hope that users would stop using Central Facility's DNS validators. Of course, besides the inevitable non-response by almost users, some users would probably notice, figure it out, and care. But as always, enough of the bosses and their minions won't believe or care. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Geoff Huston on DNS-over-TCP-only study.
From: Geoff Huston g...@apnic.net On the other hand its no more serious than any other form of small TCP transaction based services that are subjected to massive volumes, such as, say, a search engine front end. Isn't that why HTTP, SMTP, and other TCP transaction services have been changed to reduce the ratio of TCP connections to transactions? Isn't it also true that DNS transactions are much lighter weight than HTTP, SMTP, ando other TCP transaction applications? Could the gTLD roots exist in anything like their current forms if DNS transactions cost as many CPU and stable storage computrons as an HTTP GET of a purely static page (even without TLS)? Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DNSResolvers.com will be shutdown
From: Feng He fen...@nsbeta.info http://blog.easydns.org/2013/06/27/dnsresolvers-open-resolvers-will-be-shut-down/ The DNSResolvers.com free and open public resolvers will be shut down, Sorry to hear that but we have met the same DDoS problem days ago so we have to stop the free DNS hosting. Just for my curiosity and not to suggest changing any plans, were those attacks intended to hurt the open resolvers themselves or were they reflection attacks? In either case, was RRL tried? Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Clear DNS cache
..It seems your nameservers don't agree on the SOA serial number!... I wouldn't put too much stock in what http://viewdns.info/ says about anything, and not just because what how third parties digest your RRs is not dispositives or because historically the web DNS digesters have always spread a lot of bogus fear, uncertainty, doubt, and misinformation. All that really matters is what `dig`, `nslookup`, other tools, and recursive and stubb resolvers say. They're badly confused about the DNS RRs for rhyolite.com. Never mind what I suspect are their glue confusions, perhaps due to IPv6 or perhaps due to my using well distributed secondaries. Besides your nameservers don't agree on the SOA serial number they also say this about my SOA: Your Start of Authority (SOA) record is: Primary nameserver: 5 Hostmaster E-mail address: 2 Serial number: 28800 Refresh: 20130815213614 Retry: 20130616213614 Expire: 26805 Minimum TTL: rhyolite.com. and then hector me about the implications of that silly nonsense. This is what an old version of `dig +dnssec` on someone's system (not mine) says: rhyolite.com. 27587 IN SOA ns.rhyolite.com. named-mgr.rhyolite.com. 1371422174 3600 900 2592000 7200 rhyolite.com. 27587 IN RRSIG SOA 5 2 28800 20130815213614 20130616213614 26805 rhyolite.com. uTprgMR4QbNDzyBKCgDUINT1ToLVnSvB9UZ3IOoNofQmx9kQ5u8toMj+ aEX+MN7cUJqyXvYqrG3f4jf9ezfXEaOUkaMVGYitXK+FfA80jOGL2d9s EPSGjFrPu47mcy8hbkz9PAYtMY1wG/4iIpy/kJLXB/sRMfkdwtA7NKst s0M= Notice the 20130815213614 in the RRSIG. I think an exegesis of RRs by code written by someone who didn't reflexively deal getting unexpected RRs from strange DNS servers should not be interesting to anyone, and especially not when the extra RR is standard and only included when you explicitly ask for it with the flag bit. They also say: Your Mail eXchanger (MX) records are: 5 2 [TTL=IN] and they point out the various crazinesses of that. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Best Practices
From: Florian Weimer f...@deneb.enyo.de a) Secure configuration guidelines (RRL you can't make part of that, because it requires too much tuning IMHO). I think RRL is young for published, official secure configuration guidelines. rrl's defaults work fine on every authority server i've tried. That's probably because those servers don't see traffic from resolvers which in turn have clients that send queries which are a little bit creative. What is the nature of the troublesome traffic or tuning on authority servers that you've seen or heard about with settings as close to defaults as you get without leaving RRL turned off? rate-limit { responses-per-second X; }; probably for X=2 and =20, and perhaps X=15 as suggested on http://www.redbarn.org/dns/ratelimits Would the too much tuning problem be fixed by adding rate-limit { responses-per-second 15; }; to the example in the BIND9 ARM text? ISC-TN-2012-1 is unfortunately not very clear about the actual key used to determine the bucket to account against. What is the relevance of the shortcomings of http://ss.vix.su/~vixie/isc-tn-2012-1.txt to whether RRL works on authority (or even recursive**) servers without too much tuning? Section 2.2.1 claims that many possible questions can yield the same answer and suggests that the rate limit applies to those same answers (which apparently do not include the transaction ID or question section), but section 3.1 talks about the QNAME. It wouldn't make sense to rate limit based on transaction IDs, because they're supposed to be functionally unique. The R in RRL stands for response, and so rate limits should ignore the question section as much as possible. For non-empty, non-error, non-wildcard generated, non-referral responses, the key is {class,qname,qtype,client IP block}. Section 2.2.1 is about the special cases where answers are too similar. The rate limit for NXDOMAIN responsess is applied to the domain from the SOA, because response rate limiting would not be a useful DNS reflection attack mitigation mechanism if it treated the identical responses to the practically infinite number of different random.example.com questions the same. Is there a specific question about the key in BIND9 RRL? ** I continue to be surprised and disappointed by people who spin RRL is not recommended for recursives into RRL-doesn't-work-especially-on-recursives and then flog that FUD. Not recommend differs from doesn't work, denies all service, is a security hole, or breaks the intertubes. RRL on recursives could in theory slow down applications that repeat requests a lot, but I do not recall hearing of even one case where end users noticed or complained. Recursive servers should generally not need RRL, because they shouldn't be open and so needn't worry about reflection DDoS attacks. Not recommending RRL on recursive servers is like not prescribing statins for people without high cholesterol levels. Besides, open recursives must have some kind rate limiting, as people who run professional open recursive servers say. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Best Practices
none of the retransmissions, because the first burst of 1000 drove the count negative and we're still in the window. It will slip half and drop other half. 5. repeat from #3 For 30 concurrently, simplistically rendered IMG tags, there will be 1 timeout. If your web browser limits itself to fewer than 15 concurrent connections to any single HTTP server, it probably won't be affected at all. For 1000 concurrent spam, wouldn't it be wiser to rate limit your spammers? Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DNSCrypt.
From: Ken A k...@pacific.net What is keeping nameserver vendors from building this into servers? http://www.opendns.com/technology/dnscrypt/ http://dnscrypt.org/ https://github.com/Cofyc/dnscrypt-wrapper Preventing men in the middle raises key distribution questions, so I went looking for answers. https://github.com/jedisct1/dnscrypt-proxy/blob/master/TECHNOTES says The following information has to be provided to the proxy: - The provider name (defaults to 2.dnscrypt-cert.opendns.com.) - The provider public key (defaults to the current one for OpenDNS). ... The proxy doesn't cache replies. Neither does it perform any DNSSEC validation yet. This is better handled by a separate process or by linking libunbound. That makes the proxy+stub resolver and all of its implicit practical problems sound a lot like a stub resolver with a copy of the root DNSSEC key doing its own DNSSEC validation. I see commercial advantages in this mechanism to OpenDNS in locking in customers. (I intend no offense to OpenDNS. Every business does and should look for ways to retain customers.) I don't understand why DNSCrypt is better than stub resolvers that do DNSSEC validation. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DNSCrypt.
Yes, except that DNS-over-TCP helps reduce the risk of MITM, which is a perceived channel-validation benefit of DNSSEC. How does DNS/TCP reduce MITM risks enough to talk about? How is DNS/TCP a problem for governments and other bad actors? 25 years ago I naively assumed that transparent and translucent proxies for popular TCP based protocols were not practical at scale. Then AOL started proxying port 25 and now everyone has man in the middle proxies for all kinds of TCP applications including some that are ostensibly protected with TLS. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] bind-9.9.3rc2 ANY+TCP patch
From: Matthijs Mekking matth...@nlnetlabs.nl https://indico.dns-oarc.net/indico/getFile.py/access?contribId=4resId=0materialId=slidesconfId=0 Page #12 I also wonder about the definition of false positive. There are many plausible candidates. I agree. Basically it is a query from an attacker that is not being dropped. That sounds like a false negative instead of a false positive. A false positive would be dropping or slipping a legitimate or non-attack query. A true positive is correctly identifying and dealing with an attack packet. A true negative is correctly identifying and not harming a non-attack packet. https://www.google.com/search?q=false+positive http://www.mathsisfun.com/data/probability-false-negatives-positives.html https://en.wikipedia.org/wiki/Type_I_and_type_II_errors Perhaps Slip 1 on page #12 refers to the RRL parameter. If I assume False positives means truncated responses (which are true positives instead of false positives), then all of the table except the TCP responses column makes sense. I have no idea what TCP response column means. I know it has more to it than that. It might be a good idea to define the term in the technical note. I can write some initial text, if that is appreciated. I would a appreciate a few words here. I don't understand the graphs and tables, but I agree with the conclusions on page #20. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] bind-9.9.3rc2 ANY+TCP patch
From: Matthijs Mekking matth...@nlnetlabs.nl https://www.google.com/search?q=false+positive http://www.mathsisfun.com/data/probability-false-negatives-positives.html https://en.wikipedia.org/wiki/Type_I_and_type_II_errors So a false positive is a type I error, aka the incorrect rejection of a true. True is an adjective instead of a noun in this context. The nouns in this context are postive and negative. Putting that back in RRL perspective, I translate that to a false positive is the failure to identify and deal with an attack packet (like above). That is mistaken. We are talking about testing for (and perhaps mitigating) attack packets. A positive for our test is this packet is an attack packet. Deciding that a packet is not an attack packet is a negative. An accurate test or determination that a packet is or is not an attack packet is a true positive or true negative. An inaccurate determination by the test is a false positive or false negative. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] bind-9.9.3rc2 ANY+TCP patch
From: Jared Mauch ja...@puck.nether.net I thought I'd share this to anyone that wants to just force all TYPE=ANY queries over TCP to prevent those from coming from spoofed locations. This is a crude but effective hack. It doesn't stop the system from recursing to find the response. http://puck.nether.net/~jared/bind-9.9.3rc2-tcp-any.patch I can understand simplistic DNS reflection mitigation in firewalls, especially when response rate limiting is not available in the DNS server implementation or when local policies forbid the use of patches. I don't understand why would one use a patch like that with its limitations and drawbacks (e.g. usable only on recent versions of BIND9, affects only ANY, affects all ANY, doesn't limit the flood of reflected truncated responses during attacks, no whitelisting for local clients, not view-specific) instead of the full blown RRL patch for 9.9.3rc2, 9.9.2, 9.9.2-P1, 9.9.2-P2, 9.8.4-P2, 9.8.4-P1, or 9.8.5rc2. By the way, why use qtype == 255 instead of qtype == dns_rdatatype_any ? Why #define TCP_CLIENT() and use the macro exactly once instead something like if (qtype == dns_rdatatype_any (client-attributes NS_CLIENTATTR_TCP) != 0) { If TCP_CLIENT() is used in query.c, then its definition should be moved from client.c to bin/named/include/named/client.h and the several uses of client-attributes NS_CLIENTATTR_TCP in query.c replaced with TCP_CLIENT(). It's bad form to define macros (or much of anything) more than once, because you can be sure that eventually the definitions will differ. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] bind-9.9.3rc2 ANY+TCP patch
From: Jared Mauch ja...@puck.nether.net Because of the FP ratio presented at the DNS-OARC meeting this past week. It's suitable on a recursive resolver, where RRL is most effective on an authority. See https://indico.dns-oarc.net/indico/getFile.py/access?contribId=4resId=0materialId=slidesconfId=0 Page #12 I wonder to which RRL implemetation those numbers apply? Please recall that those slides appeaer to be from NLnet Labs and that one of my concerns with the NLnet Labs RRL implementation is the possibility of significantly more false positives than what I hope are the practically none from the BIND9 RRL code. I also wonder about the definition of false positive. There are many plausible candidates. This effectively does slip=1 and does away with any amplification and just makes it a pure reflection attack. Still not ideal, but doesn't amplify. On the contrary, as I just now wrote in the ratelimits mailing list http://lists.redbarn.org/mailman/listinfo/ratelimits your patch does not affect amplification by authorities. For example, if applied to an authority for isc.org, `dig +dnssec isc.org any @ams.sns-pb.isc.org' would still reflect almost 4 KBytes for each 60 byte ANY request. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] bind force qtype=ANY to TCP
From: Jared Mauch ja...@puck.nether.net The folks that are most concerned with RRL are those expecting queries from stub resolvers, I think this would mitigate this risk. } Is it intentional that the patch does not affect authoritative ANY } responses? I think the patch would fail to stop the authorities for } isc.org from answering `dig +dnssec isc.org any @ams.sns-pb.isc.org' } with almost 4 Kbytes. } } It's somewhat accidental, but I think OK. We disagree on both of those issues. Reflections from recursive servers are bad, but reflections from authorities are as bad if only because many authorities have more resources and so can blast more bits at a DoS target than many recursives. There's also the idea that open recursives should be closed for more reasons than complicity in reflection DoS attacks but authorities cannot be closed. } I think it is fine as it primes the cache if it's a real query, but if it's } fake then it just keeps sending TC=1 until the TTL expires. What are fake and real queries? I didn't think we were talking about queries that get NXDOMAIN responses or randomexample.com. There would be no need for any patches if there were a way to distinguish forged DoS queries from real queries from the DoS target. That reference to cache priming suggested another thought. As you wrote, the patch does not stop recursion from filling the local cache. The patched code is not used when the local cache already has the answer, as it will after an initial TC=1 response, because BIND sort of pretends that it is authoritative for everything in the cache. That implies the patch should have no effect after an initial ANY query and TC=1 response. I think the patch has a false negative rate of approximately 100%. To check whether I am wrong again, I set up a test server and tried two `dig +ignore isc.org any` commands. The first got a TC=1 error response as expected. The second command got 3500 bytes of RRs via UDP. I expect (but haven't tested) that all subsequent queries get normal responses until all of the TTLs expire. So I recommend that those who want to answer all UDP ANY responses with TC=1 and don't like my real recommendation of Don't Do That! use one of the fancy iptables or other firewall rules for doing that. Or am I wrong again and no one has offered such rules?--if so, use one of the rules that simply block ANY (which I also don't like). Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DNS Issue
From: Jared Mauch ja...@puck.nether.net Because someone told them the wrong thing and they don't know any difference. Just because they're an auditor doesn't mean they are clued. Simple thing would be to show them a dns query that requires tcp, such as: Would you show anything to a doctor prescribing bloodletting to cure what ails you or would you quietly leave? (except for lab work) Someone who let a financial auditor with equivalent ignorance about the fundamentals of bookkeeping near the company's books (not to mention hiring) would fear being fired or indicted as an accessory. If your boss or boss' boss' boss etc. hired an equivalent to audit the company books, you'd infer the worst and start looking for a new job while the banks are still cashing your paychecks. The same should apply to network security quacks. Bogus security audits or auditors might not signal as much about your paychecks as bogus financial audits, but they do signal coming security disasters that probably won't help your career. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Having doubts about BCP38 solving the ORN problem
From: Fred Morris m3...@m3047.net The premise with regard to BCP38 + open resolvers is that the spoofed packets reside on different networks than the resolvers. If these resolvers are primarily CPE and other unmaintained equipment, then it stands to reason that they reside inside networks containing other equipment; and this equipment could be the source of the source-spoofed (DNS) packets. Reflecting traffic off of an open resolver on one's own network would serve to cloak the true identity of the originator. Other responses have been good about the general issue of where ingress filtering must be done, but I think that scenario is so specific that it wants a narrow response. The rest of the Internet does not care which boxes inside the customer network send the stream of UDP/53 packets. It doesn't matter to the rest of the Internet whether the bad guy reflects them off resolvers inside the customers network or sends them directly. On the outside, it *is* a simple DoS attack without reflection complications. The customer might care, because floods reflected off internal resolvers might cause spikes on the resolvers that might be more noticable than valid packets (at least not violating BCP38 filters) directly from the corrupt systems to the outside target. Floods sent directly from the corrupt systems to the distant target instead of by reflection will have fewer packet losses. If the bad guy can forge source addresses, then a Dos attack direct from the corrupt system can appear to come from the customer's resolver, but without triggering rate limits or other defenses on the resolver. Only if the customer has unusual firewall rules limiting *outbound* UDP/53 to the resolvers and if those rules are not based on more than IP source address would reflections be better for the bad guy than sending directly. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Force TCP for external quereis to Open Resolvers?
From: Xun Fan xun...@isi.edu to discuss here is TCP. Someone says TCP is expensive, but if we could afford entirely shutting down external queries, then two more RTTs to get a response seems trivial. clientserver 1. DNS request/UDP -- 2. -- DNS response/UDP A normal DNS transaction would end here. Forcing TCP requires the following: 3. TCP SYN -- 4. -- TCP SYN-ACK 5. TCP SYN-ACK -- 6. DNS request/TCP-- 7. -- DNS response/TCP 8. TCP FIN -- 9. -- TCP ACK 10. -- TCP FIN 11. TCP ACK -- (That's what I see with `tcpdump -n -i ZZ0 port 53 and host XXX` during `dig +vc .com @XXX` Try it yourself.) That increase from 2 to 11 packets and from 1 to 5 round trips is not the only cost. There is also dealing with the pile of transmission control blocks (TCBs) for the duration of the time-wait delay, and on a busy server those costs can be worse. . And as a internet measurement researcher, I also find the value of open resolvers in some research projects that OR greatly extend our view to the Internet. I would like to find a way to solve the problem that we are facing now, while preserve the open resolvers for its good side. Open resolvers are not certainly not justified by the needs of researchers. So do you think force TCP for external queries to OR is a feasible solution to DNS reflect amplification problem? There are several reasons why it is not feasible. The owners of almost all of the many millions of open resolvers would be happier if they were closed. Almost all open resolvers are unintentionally open and use of them by outsiders is an objectionable waste bandwidth, CPU cycles, and other resources. It would be easier for their owners to close them than to change them to force DNS/TCP, because in many cases closing consists of correcting configuration errors or adjusting firewalls to drop incoming packets addressed to UDP/53 while forcing TCP requires changing software (or CPE firmware). Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Force TCP for external quereis to Open Resolvers?
From: Jim Reid j...@rfc1035.com I'm not sure it will make a difference though. The bad guys won't bother to do TCP for the obvious reason and will stick with their current, DNS protocol conformant, behaviour. The bad guys would not be able to stick with anything. The idea is to change all DNS servers to answer all DNS/UDP requests (or perhaps all outside requests) with truncated (TC=1) responses to force clients to retry with DNS/TCP. It might make sense for the few resolvers whose owners want them to be open (never mind that most of those owners are mistaken), but it assumes that it is possible to install new software on 21 million open resolvers that are open only because they are not properly maintained. Remember too that in these DDoS attacks truncated UDP responses would still be going to spoofed addresses. So those victims still get hit, albeit without the amplification factor of a chubby DNS response. That amplification is the reason why the bad guys bother. I expect TCP to an anycast resolver -- say 8.8.8.8? -- will prove tricky for long-lived connections. Which long-lived DNS/TCP connections are those? DNS/TCP to 8.8.8.8 in `dig +vc example.com @8.8.8.8` works for me and takes a fraction of a second. (`dig` claims Query time: 50 msec, but that evidently only covers one of the TCP round trips. `tcpdump` timestamps show a total of about 150 ms.) Keeping state for bazillions of DNS TCP connections to a resolving server will present further challenges. Yes, that could be a problem on busy DNS servers handling lots of legitimate traffic. The costs are not only holding the TCBs for the fraction of a second of a DNS/TCP transaction but holding them for the time-wait delay. See https://www.google.com/search?q=tcp+time+wait+exhaustion [Maybe TCPCT could help.] I don't see anything in https://tools.ietf.org/html/rfc6013 that reduces the costs of TCP for DNS. Perhaps you mean T/TCP to bypass the TCP 3-way handshake. However, the expensive 3-way handshake in which the DNS client says Yes, I really did send that DNS request is why DNS/TCP prevents reflection DoS attacks. If you bypass the 3-way handshake, you get the same reflection DoS tool that you have with DNS/UDP. If you can change the software on 21 million open resolvers to use DNS over T/TCP, why do the the easier thing of closing them? If you could change their software and want to keep them open, then you could install RRL and DNS cookies. The problems that RRL has on resolvers would be solved with DNS cookies. DNS cookies don't need kernel changes but only changes in DNS client and server software. https://tools.ietf.org/html/draft-eastlake-dnsext-cookies-03 Another problem is lots of crapware -- CPE, hotel networks, coffee shop wi-fi, etc -- assume DNS is only ever done over UDP. That invalidates many rationalizations for keeping resolvers open. In real life, travelers wanting to use the home office resolver must use VPNs and so don't need the home office resolver to be open to outsiders. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Force TCP for external quereis to Open Resolvers?
Only the DNS people think that. The HTTP people are used to many TCP connections to manage and do not think it is impossible. So we could abandon DNS/UDP and move exclusively to DNS/TCP? No one said that it is impossible to handle lots of DNS/TCP connections. It is a simple, unavoidable fact that TCP is far more expensive than than UDP not only in bandwidth, latency, and CPU cycles but also memory. I spent years whacking on network code at a vendor once known for high network performance to improve HTTP hit numbers as well as UDP and TCP bandwidth numbers. There are many things that you can do to speed up DNS/TCP, but DNS/UDP will always be a *lot* cheaper. Switching from DNS/UDP to DNS/TCP requires more memory, CPU cycles, and bandwidth. That's obviously not impossible, but it's also not free. If you could change the 21 million open resolvers and for crazy reasons wanted to keep them open, there are cheaper ways to make them useless for reflection attacks than the TC=1 hack. But if you could change them, you would close them for simple hygene and so not care about DNS/TCP, T/TCP, DNS cookies, or anything else. In the real world, the only hopes for fixing the 21 million open resolvers are - protecting them with BCP 38 (faint) - blacklisting them at enough authoritative servers and so forcing their owners to wake up and do something (also faint). - firewalls at ISPs filtering incoming UDP/53 (I'm not holding my breath, since that's similar to the work of BCP 38) - scanning for them and nagging their owners with unsolicited bulk email or spam (hopeless as demonstrated with SMTP relays) - years and years and years and years of patience .. } From: Jim Reid j...@rfc1035.com } In this case, DDoS attackers would get those truncated responses } sent to their victims. OK, they lose the amplification factor but } they still get to flood the victim(s) with unsolicited traffic. If } that lost payload matters to the attacker, they can just ramp up } the size of their botnet or the number of reflecting name servers } to compensate: Without amplification by reflection DNS servers, the bad guys can deliver more bits/sec at their targets by sending directly to the targets. Bouncing bits off mirrors that don't amplify results in fewer bits at the targets as some packets are inevitably lost. What's the profit for the bad guy in spending 10 bps of botnet bandwidth to reflect 9 bps at the target? Bad guys that send from a few sources instead of a botnet might hope to hide behind DNS refelections, but to hit a target with 300 Gbps they'd need to send more than 300 Gbps from those few sources. Tools for detecting and tracing and then terminating such large streams exist and are being improved. } I expect TCP to an anycast resolver -- say 8.8.8.8? -- will prove } tricky for long-lived connections. } } Which long-lived DNS/TCP connections are those? } } I was thinking of the use case where an application's resolver } opens a TCP connection and assumes it stays open until the application } goes away: eg the resolver in a web browser opening a connection } to 8.8.8.8 and shoving all its DNS lookups down that until the web } session ends some hours later. Let's accept the unsupported assumption there are any lived DNS/TCP connections in the real Internet. (AXFR and IXFR are irrelevant here.) Many things break long lived TCP connections. If the client software is not lame, stupid, and written by idiots, it does the obvious, standard, and trivial. When write(), send(), sendmsg(), or whatever reports that the connection died, reasonable TCP client software makes a new connection. For HTTP, SMTP, and other applications reuse TCP connections to save the CPU cycles, bandwidth, and latency of the 3-way handshake and application authentication, this is not theoretical. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Force TCP for external queries to Open Resolvers?
From: Paul Wouters p...@nohats.ca Not all open resolvers are run by brainless admins. And I believe open resolvers are crucial to the open nature of the internet. There is a much better case for open SMTP relays, but we all know how that turned out. More power to you if you can follow the lead of Google, OpenDNS, and others in running open resolvers that do not abuse the rest of the Internet. However, in real life the history of SMTP relays is being repeated. Not only are almost all open resolvers orphans that would be closed if their owners knew about them, but most intentionally open resolvers are run by brainless admins with silly delusions of being competent enough to prevent abuse. If you (anyone) are running an open resolver and not deluded, then great!--but just as with SMTP relays, if you are running an open resolver or relay, you're probably fooling yourself. ] From: Joe Abley jab...@hopcount.ca ] There seems to be an implicit assumption in this thread that when ] we say DNS over TCP, we mean setting up a TCP session and tearing it ] down again once per query. In practice on the real Internet, that is what will continue to be so for the forseeable future. If we could change those 21 million open resolvers to cache TCP sessions, then we'd also close them and so not need to pay any of the costs of TCP. ] If instead we imagine persistent pools of TCP connections open between ] stubs and resolvers which are rarely set up or torn down, how is the ] overhead in bandwidth, latency and CPU cycles substantially different ] from UDP? For the duration of the TCP connection, you use only 3 packets per request (request, response, and ack unless the ack is piggybacked on the next request) and so only 50% more bandwidth than UDP. However, even if you don't think 50% more bandwidth and packets matter, there are cheaper ways to save enough state to recognize repeat clients. Neither the client nor the server need a 100 or 200 byte TCB for DNS cookies. https://tools.ietf.org/html/draft-eastlake-dnsext-cookies-03 With DNS cookies, servers don't need to save any state at all. That sounds better than expecting the roots to maintain millions of open TCP connections. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Force TCP for external quereis to Open Resolvers?
From: Xun Fan xun...@isi.edu What we discuss here is for those administrators who are willing to do something to their OR. Look at what options they have now: 1) keep open = DNS amp attackers are happy 2) close = no one can query from outside The idea that those are the only alternatives is as mistaken as the idea that DNS/UDP packets forcing TCP would contain 512 bytes. You could invalidate the idea that those are the only current alternatives by noticing that Google's 8.8.8.8 is open, famous for a long time, and not abused. I've recently seen more than one reference to https://developers.google.com/speed/public-dns/docs/security#rate_limit You could invalidate the idea about 512 byte truncated packets by looking at the last line of `dig` output or using wireshark, tcpdump, etc., by simply understanding the DNS protocol, or by reading many of the places where forcing TCP with TC=1 has been proposed. To get an example of a truncated response, provoke one from a server that uses RRL and a non-zero SLIP value with `repeat 50 dig ...`. Please read http://www.redbarn.org/dns/ratelimits and the pages linked from there for another supposed panacea for intentionally open resolvers that is not as obviously broken as TC=1 forcing TCP. When you understand why RRL is not a general solution for open resolvers (not to mention noticing that RRL includes using TC=1), perhaps you will also see why TC=1 is not a solution for intentionally open resolvers. (That some people have reported good enough results with RRL on their open resolvers does not redeem it for general use on open resolvers.) Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Force TCP for external quereis to Open Resolvers?
From: Paul Vixie p...@redbarn.org also, in TCPCT there's room for a payload in the SYN. In theory there was also room for a payload in the TCP SYN before popular defenses against syn-flooding. in practice this means a normal three way handshake for the first connection between an endpoint-pair, but there's a single round trip on any subsequent connection between that endpoint-pair, involving one packet to send the request, and one or more packets to send the response. level -- i think tcp/80 could benefit from zero state cost in responders, and single round trip for request plus multipacket response, http://static.usenix.org/publications/login/2009-12/openpdfs/metzger.pdf. argue for TCPCT i'm arguing for it on the general principle that we'd like a responder to have proof of requester identity before sending a multipacket response. we would not use these powers to make OR ubiquitous. That bit about mult-packet responses is critical. Replacing 2 DNS/UDP packets with 9 DNS/TCP or 9 DNS/TCPCT for an isolated request is unprofitable. However, if the DNS response is not a single =512 byte UDP packet but a train of DNS/UDP/IP fragments carrying 2 or 3 KBytes, ... Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] FYI: SAC057 - SSAC Advisory on Internal Name Certificates
i certainly hope the reference to hr being a local or internal or non-unique name is a mistake and that CAs would absolutely refuse to issue certs for names that are the same as a really existing TLD: Not using FQDNs is foolish and unwarranted - and issuing certificates to match unqualified names is not improving the general picture. What I find more disturbing is this: ... What I find even more disturbing is that people are still talking about commercial PKI as if it it had been other than expensive security theater for more than 10 years (i.e. since CA-2001-04). Instead of spending effort on equivalents to arguing that a black semi-automatic rifle is too dangerous for civilians but the same weapon painted pink is ok, or that a molded hand grip makes a 2 inch knife too dangerous for an airplane, spend it on things with at least a little real world security implications such as DNSSEC and eventually DANE. Don't waste time lobbying ICANN, but do urge browser vendors to start using TLSA records. It might be extreme and it's certainly unitentially offensive, but a case can be made that no one writing from a domain without RRSIGs on its MX and A RRs should say anything in public about network security other than to ask about DNSSEC. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Odd MX queries
From: Daniel Stirnimann daniel.stirnim...@switch.ch I'm using the current BIND9 9.8.4 RPZ+RRL patch. It's completely evading DNS-RRL on the tld-nameserver where a lot of different query-names and the RCODE is NOERROR. All of the domains in the first list in your previous message give me NXDOMAIN. How is it evading the the BIND9 RRL referral limit on your TLD server? On the 2nd-level name-server the MX query rate is only about 120 qps. I guess it's too few queries to trigger my generous DNS-RRL config. I have response-per-second 20. For example, within 15 minutes 81 different query-names are sent. The domain which is queried the most is used 186 times within 15 minutes. That's way below the DNS-RRL config threshold. However, it's nothing which concerns me. As said, the abusive traffic on the 2nd-level names-server is quite low. On the tld name-server it was different. Yes, 81 names/15 minutes is only about 0.1 qps. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Odd MX queries
From: Daniel Stirnimann daniel.stirnim...@switch.ch One error I made is that there are lots of different IP addresses sending these queries. The IP address 203.45.217.122 which I referred to in my original post sends about 50 qps but there are roughly 5800 other IPs sending this traffic as well. Some only one query within 15 minute but most something between 1 qps and 40 qps. That's interesting. The few IP addresses which send more then my threshold (response-per-second 20) are rate-limited. That's a relief. If I were eager to repeat the very popular error of confusing guesses with knowledge and facts, I might expound on botnets and spam and claim that the increase in spam backscatter in my personal mailbox and the ~7% increase in spam reported to DCC are both real and related to what you are seeing. http://www.rhyolite.com/dcc/graphs/?BIG=1end=1363032000resol=1m http://www.rhyolite.com/dcc/graphs/?resol=1wend=1363032000BIG=1 http://www.rhyolite.com/dcc/graphs/?resol=1wend=1361822400BIG=1 However, I've learned from many years of watching others make authoritative sounding declarations about the what, where, why, and how of network evil, and be immediately or sooner shown to be full of negative clues (facts that are false). Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Another whitepaper on DDOS
From: Tony Finch d...@dotat.at But the errornous transfer of ebay.de would create a deasaster with DANE. In what way would DANE make the theft of a domain worse? In addition to vjs's points, note that DNSSEC makes theft of a domain even more visible because it is likely to cause horrible breakage for validating users. I didn't mention those alarms, because I assumed the domain was stolen at the registrar or in the registry so that glue and DS records would be corrected by the adversary. I didn't recall the particular theft, but assumed it involved the common modes of seizure by the registrar or the use of stolen credentials at the registrar. Only if the theft is downstream of the registry such as in a master authoritative server for the domain would DNSSEC raise alarms. Those alarms are valuable, but I didn't want to argue nits with people who after much more than a decade and many public scandles, still haven't twigged to the unredeemable fraud that is commercial PKI. Never mind the irony in the likely fact that the use of stolen registrar credentials would be protected (sic) by commercial PKI. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Another whitepaper on DDOS
I wonder if DANE could have prevented Microsoft's recent difficulty with expired SSL certs. https://www.google.com/search?tbm=nwsas_q=microsoft+azure+ssl Instead of an annual bout with internal purchase order and invoice red tape and with red tape at the CA, could Microsoft have automated the generation of certs and fingerprint TLSA RRs just as many automate their generation of zone signing RRSIG RRs? (Never mind that microsoft.com lacks RRSIG RRs.) ... From: Doug Barton do...@dougbarton.us Are there CA vendors who give out EV certificates for $fee + answer the e-mail? I know you can get basic SSL certs simply by answering the e-mail from the CA. I can't find anything about EV verification from registrars. Maybe I'm blind and stupid, or maybe writing down what they actually do would be too funny. I suspect you might need to submit a government registration document and answer a press-1-if-you're-human robo phone call. You won't forge the registration document, because the real things are so cheap, easy, and unverified. It's obvious nothing that I put in the online form other to get http://www.sos.state.co.us/biz/BuildCertificate.do?masterFileId=20051118531 was verified other than the credit card number for the $1.00 charge. (You might need to 'get' that URL twice.) See also http://www.sos.state.co.us/pubs/info_center/fees/business.html I've had DBA registrations in other states, and found them just as unimpressive. How would you interpret section 5 of https://www.cabforum.org/EV_Certificate_Guidelines.pdf to sell me a $1500 EV cert? https://www.symantec.com/theme.jsp?themeid=compare-ssl-certificates You couldn't afford to have someone to drive past my address to see if it's a vacant lot, not to mention ask my neighbors if they've seen anything shady or even ever seen me. If you want to sell certs to small businesses, then you cannot charge enough to do any checking. Not that look for the green bar is going to be a whole lot more successful than Don't say yes to security exceptions you don't understand, but I'm curious. :) Yes, EV certs are expensive tickets for slapstick security theater. Standards certs and the mailboxes (not SMTP but only for use after you log into your GoDaddy account), theft protection, scanning, and other hookum that GoDaddy sells are cheap seats. (Your recent claim that all registrars up-sell the same junk as GoDaddy is wrong. I'm trust that all of the registrars you've seen are as you said and like GoDaddy, but I've seen nothing like GoDaddy. That might be because I don't look at registrars that I've heard bad things about or that advertise prices below what I know of their costs (e.g. registry fees). I know they'll more than make up their losses in ways I'm too dumb to catch.) Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Defending against DNS reflection amplification attacks
From: Joe Abley jab...@hopcount.ca If you can describe BCP38 deployment in a non-trivial network such that deployment is to the benefit of shareholders and non-deployment is not, I'm all ears. Absent regulation and punitive fines for non-compliance, I don't see it. Civil lawsuits by victims of DNS reflection and other attacks that depend on failures to deploy BCP38 might help convince boards of directors. It might help to take up a collection to help pay the legal fees a victim sueing one of those non-trivial networks. I've the vague impression that kind of fund raising is illegal. I've learned to avoid using the word fine in a different but related context. I have long claimed that ESPs (bulk mailer for hire) could practically stop the large amounts of unsolicited bulk email that they send by fining their customers with dirty target lists. A $100 fine for each spam complaint verified by the ESP (maybe only after the 5th complaint and maybe capped at $5,000) would practically stop the ESP spam sent toward my personal mailbox and to my spam traps feeding DCC. A representative of a major ESP insisted in public that my claim is nonsense, because it is illegal (sic) for an ESP to fine its customers. Because ESPs are private enterprises, that might be literally true. It's also a lie because ESPs could say cleanup fee or spam complaint processing fee instead of fine without reducing the disincentive for purchased, harvested, re-purposed, or other dirty mailboxes in target lists. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Another whitepaper on DDOS
From: Lutz Donnerhacke l...@iks-jena.de But the errornous transfer of ebay.de would create a deasaster with DANE. In what way would DANE make the theft of a domain worse? Without DANE, the new possessor of a domain need only get SMTP working, create a new cert, apply for signature for a new cert, answer the email from the CA verifying ownership of the domain, and start using that new cert on new HTTP servers with improved web pages. With DANE, only a few things differ. One difference is that the new cert can be used as soon as DNS TTLs allow without waiting to answer ownership-verifying email from the CA. The second difference is that before and after the transfer, browser users can be more confident that the web pages they see are unchanged between HTTP server and HTTP client. In no case can you be sure that ebay.de is what you assume it is without some sort of out-of-band exchange of keys and secrets between you and ebay.de. Paying a CA $500 cannot buy more than $500 worth of identity checking and authentication, and that cannot penetrate more than $500 worth of smoke, mirrors, forged business licenses, etc. $500 is plenty for a hobby domain but ridiculous for an eBay. (Never mind the free CAs.) Commercial PKI verifications of the identities of strangers have always been frauds and snake oil sold to punters. That commercial PKI fees have always been too small to allow honest identity checks even for organizations more famous than Ebay was proven more than 10 years ago. https://www.cert.org/advisories/CA-2001-04.html http://technet.microsoft.com/en-us/security/advisory/2524375 Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Another whitepaper on DDOS
From: Jeff Wright jwri...@isc.org http://docs.media.bitpipe.com/io_10x/io_106675/item_584633/Gartner%20and%20Arbor%20Focus%20on%20DDoS%20FINAL.PDF On one hand, it - gets significant bits of history wrong, such as the claim that SSL had nothing to do with authentication and authorization until EV certificates. If confidentiality (encryption) were the sole point of SSL, then SSL would have gone straight to a DH exchange and done no public key computing. EV would not be a minor elaboration of the old, widely used PKI. (page 15) - urges the use of DNS Authentication. I guess DNS authentication [would work] to ensure that source queries to a DNS server ... are in fact coming from a valid host if you can find and deploy DNS stub resolvers that support DNS authentication and then deploy them. I think that's practically impossible for the forseeable futgure. It might instead be referring to ACLs in servers and relying on IP source addresses as authentication tokens, but that would be almost as lame. (page 6) - advocates naive and so bad query rate limiting and separate NXDOMAIN rate limiting. It should have mentioned RRL. (page 6) - advoctees applying RexEx's and packet capture for no purpose. Looking for text in DNS packets will find lots of it separated by what look like ASCII control characters. Unless you have a specific target, you're unlikely to do more than waste time by manually staring at packets for any port. (page 6) On the other hand, those are all minor nits and mostly reflect my prejudiced and overly strict reading. Overall, I found it innocuous and entertaining. If it seems revolutionary or eye opening and you have relevant responsibilities, then you urgently need more than any such document can offer. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] RRL specified in a stable place?
... an Internet-Draft followed by an RFC would be Really Helpful. What track do you expect this to go along? Is this a DNSOP draft? Those are best boring details except to those with non-technical interests in the IETF. Because the implementations are really just a way of using existing parts of the specifications in creative ways. That covers a lot of RFCs. (They're also risky for some operators. That covers a lot of RFCs. Consider that, if you spoof $ISP's resolver addresses and perform one of these attacks, then $ISP gets at least degraded service during the rate limit period. Perhaps I misunderstand, but I think that's wrong in general and based on the persistent and by now very irritating confusion between client rate limiting and response rate limiting (RRL). In addition, ISPs have reported that installing and turning on RRL has restored DNS service that had been degraded by apparent DNS refection DoS attacks. While your DNS servers are trying to respond to Mqps of bogus requests, they are often not only flooding the DoS victim but also not getting out other responses. it's not a panacea either, That cover a lot of RFCs. and certainly cannot be considered a BCP for all use cases.) Ok, so don't make as a BCP. Let it be Informational. Or don't advance it after publishing it as an I-D. Keeping change control out of the IETF will not slow the spread of the idea or harm interoperability (which in the old days was the reason for RFCs). If you think not letting the IETF improve and then freeze the specification will lead to fragmentation and disparate implementations, then you should oppose an RFC. Without an RFC, there is more room for better ideas. Because there is no directly involved on-the-wire protocol, there is much less need for an RFC. Personally, I think it would be nice if it were published at least as an I-D ensure that the idea reaches more potential implementors. However, it would not be a big deal if the IETF doesn't want it even as an I-D. The one thing the IETF really should do (if it has not become interchangable with the ISO/ITU/UN) is to add two check list items before advancing future protocols: - all servers MUST deal appropriately with excessive requests such as by rate limiting by client, network, request, and/or type of request. This is particularly important for services that do not require long lived state. - all clients MUST rate limit their requests, both retries and de novo, including using at least exponential back-off. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Monday rant againt the uses of the Public Suffix List
From: Warren Kumari war...@kumari.net Continuing the sarcasm is too much effort, so I'll simply ask why not do DNS MX and A requests? (both because of the fall-back-to-A-if-no-MX Please sir, if I run www.images.example.co.uk, can I set a cookie at images.example.co.uk? How about example.co.uk? Fine Now .co.uk? If you are running www.images.example.co.uk, then you should know all there is to know about cookies at www.images.example.co.uk any other domains at which you might legitimate want to set a cookie. If you are an HTTP client implementor, then I think you should implement disable third party cookies with the single obvious, fast, simple, and--if you like--simplistic comparision without needing to check any PSL lists. You should also make disable third party cookies on by default. Yes, I am among the many who consider third party cookies at best undesirable and generally willful and knowing attempts to sell or otherwise violate our privacy. Yes, I've occassionally encountered web pages that apparently legitimately use third party cookies (i.e. without obviously trying to violate my privacy). I cannot recall any cases where those web pages could not and should not have used other tactics. Yes, I know all HTTP server operators values my privacy. However, the values that spammers, advertisers, governments, and other snoops place on my privacy differ from mine. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] responding to spoofed ANY queries
From: Frank Bulk frnk...@iname.com Perhaps the ratio could be a dynamic whitelist -- if it's 1.5 or less, then allow the response to go out. What would be gained by spending the code complexity and CPU cycles such a mechanism would require? What bad things would be avoided or good things achieved? (Please do not mention false positives, because that notion of false positive is irrelevant and does not happen with RRL.) Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Can you force your IPv4/v6 DNS server to return v4 responses only on recursive lookups
From: Patrick, Robert (CONTR) robert.patr...@hq.doe.gov We need an option like this `break-dnssec` feature to use RPZ for stopping user access to DNSSEC-signed domains that are on a block list. How should it differ from the break-dnssec yes/no modifier for the response-policy{} statement mentioned in the ARM for BIND 9.9 and 9.8? Look for break-dnssec in http://ftp.isc.org/isc/bind9/cur/9.8/doc/arm/Bv9ARM.ch06.html or http://ftp.isc.org/isc/bind9/cur/9.9/doc/arm/Bv9ARM.ch06.html There is a single break-dnssec bit for each view. It seems likely that those who want to break DNSSEC with RPZ want to do it for the entire view. In addition, the rules precedence rules (and code) for choosing which polizy zone to apply are already too complicated without a separate break-dnssec bit for each policy zone. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] responding to spoofed ANY queries
From: David Conrad d...@virtualized.org I suspect you're misunderstanding what I'm saying ... Yes, it is yet another form of security theatre, but when has that stopped anyone? Yes, I misunderstand your position as the same as others'. However, I'm pretty sure this isn't appropriate fodder for dns-operations... Perhaps not, if the supposed lack of laws is not an excuse for DNS recursive server operators to keep them open and for authoritative servers to refuse to install some kind of rate limiting. The many years of stop bothering me, spam isn't illegal responses from operators of open SMTP relays and other spam-critical services make me wonder. There are many open recursive servers and authority servers without rate limiting or with RRL manually disabled except for the previously common flavors of attack. ] From: Frank Bulk frnk...@iname.com ] If the problem is amplification, why not only perform RRL on only those DNS ] communications exchanges that have certain amplification factor (i.e. 1.5). That sounds nice but has problems. The main one for me is that you'd have wait until the response has been marshalled before determining it size and deciding whether to drop it. That seems to me harder to code in BIND9 and more expensive in CPU cycles. A better reason is that simple A requests are much smaller than typical non-DNSSEC responses. `dig iname.com @204.74.108.1` sends 38 bytes and receives 225 for 5.9X amplification. 5X is not as flashy as 30X, but is a big problem. 5X is a lot more than your 1.5X and so in practice you would rate limit all responses. If you always do it, you might as well do it in the cheapest way possible, before knowing and regardless of the size of the response. Even 1X or no amplification could be useful to a bad guy wiggling through firewall holes or obscuring an origin. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] responding to spoofed ANY queries
From: David Conrad d...@virtualized.org The tool is too tempting and potentially effective for too many government projects ranging from national hostilities to operations by law enforcement against criminals to expect governments to entirely support BCP38 or even allow its complete deployment. This is like the prospects for governments and politicians limiting their own spam. A possibility but I've not yet reached that level of cynicism. I suspect that when there is a sufficient demonstration of the effectiveness of source address spoofing against government or infrastructure targets, laws will suddenly appear that require ISPs to take steps to ensure the traffic they source has appropriate source addresses, just as laws appeared to allow lawful intercept of traffic. Wouldn't spoofing against government or infrastructure targets invoke the Patriot Act and other terrorism laws? Would an ISP that hasn't deployed the recommend, available and official standard measures to prevent such attacks be an accomplice in a violation of the CFAA? The laws mandating support for wiretaps are in the opposite direction, because they mandate support for network abouse. Laws requiring that all routers support one or more of the BCP 38 mechanisms sound rather late and redundant and wouldn't do much to make ISPs turn them on, especially given the occassional perfectly legitimate situation where simple ingress filtering is wrong. More relevant than CALEA are anti-spam laws and the current noise about Iran being the source of recent reflection attacks. (Never mind whether that noise true this time or is merely more lies and FUD from the usual suspects and beltway bandits.) Everyone with experience in the spam realm knows how impotent the anti-spam laws have been. Even if someday one nation after all these years of broken promises really does outlaw unsolicited bulk email, there will still be plenty of others that won't. Why doesn't the same dire problem affect laws against all forms of network abuse including IP header forgery? Then there is the enforcement problem. Would you have DHS inspectors checking compliance? Would they spot check cages in data centers, consumer access routers, and so on and so forth? That sounds like a bigger job airport security. Would the inspectors be as competent, trustworthy, and educated as TSA inspectors? A common response reaction at this point is something about the civil courts. Why haven't the targets of the recent reflection attacks sued anyone? All authority servers that are not negligent should by now be doing something, whether RRL in BIND or NSP or operators standing by with axes. Reflecting recursive servers have no excuse besides desires to make money cheaply. I suspect some of the ISPs of the sources of the forged requests have been identified, but I've not heard of any court cases against ISPs. Besides the lack of action from the victims, there are the lessons of spam history. You won't find any signs of the civil legal victories of AOL and Earthlink in charts of spam volume. Unless Spamford Wallace goes down on electronic mail fraud, intentional damage to a protected computer, and criminal contempt, will he ever really retire? https://en.wikipedia.org/wiki/Sanford_Wallace IP source address forging is like spam. Not really. Spam doesn't affect anything except email. Source address spoofing can affect _anything_ on the Internet. Even if we agreed that spam affects nothing but email (we don't), we should learn the lessons of the spam war both in general and in the effectiveness of laws on such problems. That there would be fewer interests trying to water down a BCP 38 law into equivalents of CAN-SPAM is irrelevant, because most spam is and has been illegal since CAN-SPAM was signed. In the real world, the phrase covering laws against cybercrime is security theater. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] responding to spoofed ANY queries
From: Casey Deccio ca...@deccio.net I'm not familiar with other RRL behavior, but to provide some numbers for BIND's patch: All responses to rate-limited queries are truncated, but default behavior is to withhold response altogether for only 1 out of 2 such queries (50%). As Paul Vixie often says, the goals of RRL do not include stopping DNS reflection DoS attacks but: 1. attenuating reflection attacks so much that the attacker would do more damage to the victim by sending the bogus DNS requests directly to the victim 2. not letting the attacker deny DNS service to the victim by failing to answer the victims real requests. Goal #1 is achieved by attenuating or sending fewer bits toward the victim than the attacker sends to the DNS server. With slip 2, the attacker's bits are reduced by about 50%. The attacker would do twice the damage by sending toward the target instead of the reflector. Goal #2 is approached with slip hack and by rate limiting responses instead of clients. The victim's DNS requests are unaffected unless they are for the same name and type as the attacker's forged requests. Goal #2 is practically reached while the attacker avoids the major query types. If the attack uses a major type such as A, then we rely on the probability of at least one of the victim's requests getting a slip or TC=1 response. It helps that the victim need only get one response per DNS cache lifetime. depends on the query rate. The statisticians might provide a good rule of thumb for reasonable response rate given query rates, but it seems like 50% is in fact a good starting place. With slip=2 and the victim trying and retrying a total 3 times, the probability that all of the victims responses will be dropped is 0.5*0.5*0.5 = 0.125. That makes the probability that the victim will get a response despite matching the DoS flood about 88%. That's not perfect, but not bad. If it's a mail system that will retry a few times or a user at a browser that will manually retry a failed page, it gets even better. The BIND RRL patch also has an option for scaling down the slip value (which dictates response rate) in the presence of increased query rates. I haven't had time to play with it, but the idea is smart. The impression that the slip value can be scaled down using the gross qps rate comes from an error of mine the documentation. Only the real rates can be scaled. I've proposed that the 'slip' value be scaled up the qps ratio or the square of the qps ratio to keep the TC=1 responses/second rate constant. On the other hand, any reduction of TC=1 responses (i.e. increase in slip) reduces the reason to have slip. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DNS ANY requests from Amazon?
From: Stephane Bortzmeyer bortzme...@nic.fr While rate limiting by client IP address stops a reflection attack, it also turns off almost all DNS service to the client from the server. No one in his right mind limits simply by the client's IP address. People typically also use the type of the request (today, typically ANY). See my mini-HOWTO for Linux Netfilter in this thread. That tactic makes limited sense if you assume that the bad guys are too stupid to see that they can bypass ANY filters with almost as much amplification with other query types such as A. `dig +dnssec www.nic.fr @ns1.nic.fr` offers amplification of more than 25X. + ] but my point is that it works *today*, with *actual* attacks. So, it ] definitely helps but keep your eyes open, have alternative solutions ] in place and do not put all your eggs in one basket Yes, automated response rate limiting (RRL) is too small a basket to hold all of anyone's eggs. However, a static ANY filter amounts to trying to keep all of your eggs in a handkerchief. Manually maintained iptable rules are akin to hiring jugglers keep all of your eggs in the air. (nobody asks ANY isc.org in the real world, except the attackers). The common claim that no one uses ANY is so overstated that it is false. I use ANY to diagnose real problems in the real Internet. I appreciate the BIND RRL patch and it is obvious to me that we must continue the research in dDoS mitigation, but let's not drop the mitigations techniques that work *today*. (The attackers are not superhuman, they use imperfect techniques.) That's quite true, but advocating defenses such as static ANY filters or manually installed iptable rules is like avocating homeopathy in the fight against malaria. The suggested iptable tactic would be better if the phython program somehow automatically installed iptable rules. However, that might be too effective. In actual deployments, some people may be unwilling or unauthorized (corporate policy) to install unofficial patches on a production server. That's why we should not reject blindly the OS-level rate limiters (see my mini-HOWTO in this thread). A third party's iptable rules (not to mention a third party's program that generates the iptable rules) should get as much scrutiny as unofficial patches for BIND or NSD. Neither iptables hacks, a third party's phython program to generate iptables rules, nor DNS server patches (even if official) should be used lightly. The BIND patch includes a standard test suite in bin/tests/system/rrl Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DNS ANY requests from Amazon?
From: Dobbins, Roland rdobb...@arbor.net Sure, but RRL isn't the issue; it's all the rest of what 'application firewalls' do which causes them to choke. I've yet to see one which doesn't choke under even moderate DDoS, and have never seen one which implements any form of classification in a stateless or minimized-state manner. It's well known that Roland Dobbins doesn't think much of application firewalls or stateful firewalls in general. I also don't think much of application firewalls, and not only because the FUD that is much of their brochures, the never ending broken vendor promises, or the exaggerated performace. I've been grumbling since tcp wrappers first appeared that application firewalls are usually poor bandaids for stupid application security holes that could (and should) be more securely and cheaply fixed in the applications. But all of those criticisms are irrelevant to what hypothetical firewalls might do for current and foreseeable DNS security issues. That currently popular firewalls can't cope or do only stupid stuff like ANY filtering doesn't justify rejecting firewalls for reflection attacks on principle. Besides, DoS attacks on DNS servers themselves (as opposed to using DNS servers to attack others) are best handled outside in smart (e.g. sane state table management) application firewalls. It's not good for a DNS server to discard excessive (relative to the server's own resources) requests. By the time a request can be discarded by the server, too many local resources have been burned. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DNS ANY requests from Amazon?
It's starting to look like per-client-IP rate-limiting features are necessary... There is a patch available for rate-limiting inside BIND. There is also RRL code for NSD. Please note that the main thrust of the BIND and NSD rate limiting code is response rate limiting (RRL) and *NOT* per-client IP address rate limiting. Per-client rate limiting is generally the best that can be done with simple firewall rules or access control lists, but has limitations and can cause harm. While rate limiting by client IP address stops a reflection attack, it also turns off almost all DNS service to the client from the server. Temporarily denying name service to a target has long been a major part of more serious security problems than denials of service. For example, if you need to fool your target about the IP address of www.example.com, it's handy to have the several seconds of a full set of DNS client timeouts to try many DNS transaction IDs instead only the milliseconds before the real answer arrives. With RRL (especially with the slip feature), the victim of a reflection attack often sees no change in DNS services from the rate limiting server during a reflection attack. With client IP address rate limiting, the server stops answering practically all requests from the victim. The current version the BIND RRL patch does have support for per-client rate limiting, but it exists only to satisfy popular demand. Its use is a bad idea in most cases. I've said something like this before but I keep seeing claims that BIND rate limiting is harmful or bad based on the mistaken notion that it limits requests by IP addresses instead of limiting responses by {IP,qname,qtype}. The other common claim about RRL is that it is too expensive. Never mind that much bigger servers are using RRL than the servers run by people expressing that concern. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DNS ANY requests from Amazon?
From: Patrick, Robert (CONTR) robert.patr...@hq.doe.gov I don't disagree that limiting responses is a smarter tact than limiting requests, with respect to making an informed decision prior to discarding traffic. Having zone and query-type plus response data to evaluate the client hash is more information than looking only at source and destination IP address, as may be implemented at a firewall or within the O/S. Some of these data elements could also be tracked by an application-aware firewall. Yes, you could do response rate limiting (RRL) within an application aware firewall by have the firewall do almost of all of the work of your DNS server. For example, your RRL mechanism (whether in a firewall or DNS server) must count all NXDOMAIN responses to a given IP address as identical to avoid spewing GBytes/sec of big signed NXDOMAIN responses about distinct random, invalid domains. `dig +dnssec asdf1234asdf.com @a.gtld-servers.net` gives a 1K NXDOMAIN. Referrals have a similar issue. A firewall that is as DNS aware as that should not waste the computing it has done to know whether to discard the response it computed to count. If things are ok, it should go ahead and send the response. Never mind that consistency, maintenance, and other problems that always come with duplicating things, whether definitions of constants in code or the big chunks of code and data that are a modern DNS server. ... Allow administrators the freedom to set the limit to any value and/or disable the feature, but shipping the product with a smart default may be viewed as a pragmatic step forward in noise reduction. The right RRL value depends on each server's popularity. It might be reasonable to ship DNS software with a default rate limit suitable for modest servers (e.g. 5 or fewer responses/second) and expect big server operators to make adjustments. Continuing to deploy into the wild without any rate-limiting isn't the best approach long term. Neither is tolerating unnecessary open recursive servers and ignoring BCP-38. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] First experiments with DNS dampening to fight amplification attacks
From: Dobbins, Roland rdobb...@arbor.net this sounds like a new application of 'the chemical polluter business model'. There's more to it than that, though. It's important to understand that those who are purchasing and deploying network gear often are nonspecialists, and so frustrations, project delays, etc. would crop up in the customer organizations - who would then complain... but that *IS* 'the chemical polluter business model' which is also the spam problem which is also the tragedy of too many sheep on the commons. It's cheaper and easier in the short term to pollute, ignore spammers, and over graze the commons. The bosses of the shepherds, abuse desks, and refinery engineers hear only about the costs and problems of not overgrazing, terminating profitable accounts, and not polluting. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Massive DNS poisoning attacks in Brazil
From: Tony Finch d...@dotat.at To: Paul Vixie p...@redbarn.org Paul Vixie p...@redbarn.org wrote: in http://www.ietf.org/mail-archive/web/dnsext/current/msg11700.html i was thinking that we'd add send chain as an edns option, and then add I like this plan. All of those DNS tunneling, triggering, alternate port, and other varient protocol schemes for dealing with hotels and public access points attacks on DNS are either unnecessary in the long run or depend on practically no one ever using them. They are like the ad hoc schemes subscribers to this mailing list use to tunnel other protocols home. Any popular scheme that works around DNS, HTTP, ssh, etc. man-in-the-middle attacks that become popular will be blocked, proxied, or hijacked unless most users normally use tools that detect and refuse to work with men in the middle. If the browsers and stubb DNS servers of most users did DNSSEC, DANE, and HSTS, then any men in the middle will be obvious and won't be installed except for purposes that users tolerate including access point login, employment behind corporate firewalls, and living under authoritative regimes. In addition, those tunneling schemes will not unnecessary. To put it another way, if HTTP replaced IP as the Internet protocol without any real improvements in end to end security, then the censors and hijackers would apply their tools to HTTP. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Massive DNS poisoning attacks in Brazil
From: Tony Finch d...@dotat.at You are right about dicking around with port numbers and TLS or HTTP framing. However the send chain EDNS option would be a widely useful operation for validating stubs. A stub validator could perhaps send DS and DNSKEY queries for all the truncated versions of the name between the target name and the root, which it would have to do concurrently to avoid latency pain, but then it will have to iterate this to deal with CNAME and/or DNAME chains. The recursor has already done all the work so it would be nice to get all the results back in one go. That's a good point, except I can only go with somewhat useful. On http://www.cnn.com/ just now I see only www.cnn.com, i.cdn.turner.com, i2.cdn.turner.com among about 33 images and icons. Getting the DNSSEC chains for those half dozen DNS names (I probably missed some and I disable most javascript) would save only a trivial few round trips for a stub with a cache given the round trips to fetch those images (and javascript). Besides, the saved round trips would be to the nearby trusted server that should be answering within 50 millseconds and closer and faster than the CDN box serving the content, not to mention web sites not served by the CDN box. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Massive DNS poisoning attacks in Brazil
From: David Conrad d...@virtualized.org Any popular scheme that works around DNS, HTTP, ssh, etc. man-in-the-middle attacks that become popular will be blocked, proxied, or hijacked unless most users normally use tools that detect and refuse to work with men in the middle. You're assuming the MITM attacks are intentional. No, I assume only either that the men in the middle will back off if they irritate enough users or that they can be detected. (Never mind corrupt DNS registrars or registries attacking DNSSEC.) My impression is that the majority of the issues in getting EDNS0-requiring protocols to work are due to ignorance, e.g., valid DNS responses are always UDP512bytes or valid DNS types are {A,MX,SOA,NS,PTR,TXT}. If this is true, than egregious hack workarounds like using HTTP/S as a transport will solve most of the problem (not that I think this is the best solution). If DNS/TLS/HTTP became popular, then the same actors that filter DNS/UDP for 512 bytes or less common types would have the same motives to do the same to DNS/TLS/HTTP. To filter 512 bytes or RRSIG or TLSA records, you must be looking at the bits. Breaking DNS is not accidental, not even with NAT. The reasons that require or allow you to do whatever you're doing to DNS/UDP/IP would apply to DNS/TLS/HTTP if DNS/TLS/HTTP were popular. On the other hand, if many user computers have validating stubs that compute SERVFAIL for broken DNSSEC and so make gethostbyname() in applications fail, then many users will yell at hotel concierges for $15/day WiFi that doesn't work and use LTE instead of paying $15/day. Many hotels would change and allow EDNS0 after the sign-on. Employers would either do the same or point to conditions of employement. State actors would either do the same or send whiners to gulags. } From: Andrew Sullivan a...@anvilwalrusden.com } I see. So your model is that the application asks for a TLSA record, } and if it gets one then it can infer that the record also passed } validation? } How can the application be sure the resolver is } DNSSEC-aware? The important answer is the same way the application can be sure the resolver is not some other kind of malware. The trivial answer is in the API used by the application to get TLSA records. For gethostbyname(), HOST_NOT_FOUND in h_herrno is plenty good enough for the SERVFAIL that comes from failure to validate A or records. For other record types, you need either the record set, the empty record set, or a half bit of a failure flag. Applications do not now and will never care whether a failure is due to any of the myriad of reasons for getting a SERVFAIL or REFUSED DNS DNS response, including the new reason of failure to validate. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Massive DNS poisoning attacks in Brazil
From: Andrew Sullivan a...@anvilwalrusden.com I don't think this is the problem at all. The problem is that even if you can get that out at the end point (and I can, using DNSSEC Trigger), it does you no good because your application _can't tell_ what happened. If I'm a web browser programmer, I want to be able to know whether the DNSSEC validation worked before I start using the TLSA record. Today, that is too much work (and probably reduces to implement a resolver in the browser). Browsers are certainly not the only application, even if it is true as Paul Vixie recently said that the Internet is little more than the web for most connected computers (e.g. phones). Writing DNSSEC validation code for every application that depends on accurate DNS data would be as crazy as not using libraries and daemons for other local authentication and authorization. The only reasonable solution is to give stub resolvers some of the features of recursive resolvers including DNSSEC validation and caching to make the costs of DNSSEC tolerable. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Massive DNS poisoning attacks in Brazil
From: Paul Vixie p...@redbarn.org To: David Conrad d...@virtualized.org CC: Vernon Schryver v...@rhyolite.com, dns-operations@lists.dns-oarc.net The only reasonable solution is to give stub resolvers some of the features of recursive resolvers including DNSSEC validation and caching to make the costs of DNSSEC tolerable. Why not get rid of stub resolvers completely and simply use recursive resolvers? I think the code to parse the BIND9 configuration grammar and nothing more would be excessive and grotesque.The code to support all of that stuff would be obscene. As far as only DNSSEC is concerned, you don't need a lot of the complications that a real authority server needs. (e.g. special NSEC3 database trees or lists to make big zones less slow.) Of course, if the only available code for your situation is BIND, then you could use BIND with a tiny configuration file. The package would be smaller than current Firefox binaries that send me running and screaming in horror. there's an urban legend about how the authority servers depend on caching by intermediate recursives and that if every end system had its own recursive server on board the authorities would melt. real traffic it might get the dreck percentage down to 80% but it wouldn't melt anything. No matter how over-provisioned authority servers are, I don't understand why making stubbs more like real resolvers should increase traffic to authority servers. Why couldn't you do the equivalent of moving the DNS servers named in the system's equivalent of /etc/resolv.conf to the equivalent of a BIND forwarders{} statement and putting localhost into resolv.conf? A full featured DNS server can't bypass men in the middle any more than a bare bones DNSSEC validating caching forwarder. There's no security reason to go to the real authority servers if your local DNS servers are corrupt. The bad guys who corrupted them can attack your DNS traffic going outside. All you can reliably do is detect evil, and only if you can somehow get the root key. Detecting evil is often enough of the battle. In many (but certainly not all) cases, the bad guys react to sunshine like other vampires. In the other cases, you can choose to not play the game by their rules or at all. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] First experiments with DNS dampening to fight amplification attacks
From: =?ISO-8859-1?Q?Matth=E4us_Wander?= matthaeus.wan...@uni-due.de Hmmm for authoritative servers, we might also emit a CNAME challenge.= We could encode the encrypt the correct destination in the CNAME, for A= and this is trivial. If you come back to resolve encoded-12.32.43.43.attackeddomain.com, you get 12.32.43.43 etc. There has been recently a patent granted with this method: http://www.freepatentsonline.com/8261351.html Though they don't use it do decide about blocking, Is that because converting a reflected flood of DNSSEC signed responses to a reflected flood of DNSSEC signed challenge CNAMEs is not an impressive defense for DNS reflection attacks? Never mind that packet losses during an attack can increase and so doubling the number of packets that must succeed for a legitimate DNS/UDP transaction is unlikely to be helpful. but use the CNAME challenge on every query, still providing a small amplification. This comes at the risk of running into resolver issues with NS or MX records..= and resolver CPU loads for DNSSEC signatures for all of those synthetic challenge CNAMES during an attack. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] How many kinds of DNS DoS attacks are we trying to stop ?
DNS DDoS amplifier resistance: as a thought, would it be a reasonable step, not hurting interop, to have an authoritative DNS server process a UDP-based ANY query by including, at most, an MX and any A responses in the ANSWER section and setting the TrunCated bit of the response if there were any other records skipped? Try some experiments to see if what kind of amplification you can get without ANY. I see about 20X from `dig +dnssec asadfasdf.com` If your defenses handle non-ANY attacks, then what do you gain from doing anything about in particular about ANY except more code, more bugs, more CPU cycles, and fewer queries/second? Doing any special for ANY queries makes as little sense as filtering all ICMP packets. Why is it that so much of computer security is based on the insane assumption that everyone else and especially adveraries are stupid? There is always an easy solution to every human problem--neat, plausible, and wrong. --H. L. Mencken Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] How many kinds of DNS DoS attacks are we trying to stop ?
From: Olafur Gudmundsson o...@ogud.com ... If a traffic reducer turns on TC bit in its responses, then if no TCP connection is completed during the next N seconds, the reducer can go to full drop mode. Should the DNS RRL patch stop slipping truncated (TC=1) responses if it seems that no TCP requests have been seen from the CIDR block within window seconds? pro: - it would help answer concerns about contributing to the DoS attack, because some of the slipped responses are to forged requests. - surely some DNS reflection DoS CIDR block targets lack DNS servers and the truncated responses only harm them. con: - it's not strictly necessary and might not be justified by its code and potentical bugs. - the truncated responses are infrequent and small enough that they might not matter. - small reflection DoS targets might be sending fewer than 1 request per window seconds, and so would miss the false positive mitigation effects of the truncated responses. - even large reflection DoS targets might be sending fewer than 1 request per window seconds to most DoS reflectors and so would miss the false positive mitigation effects of the truncated responses. - for obvious as well as obscure implementation reasons, the TCP seen indicator would have a few errors in the none seen direction. I've a detailed sketch of the necessary changes to the code, but I'm inclined to forget them. Opinions should probably be expressed in the RRL mailing list at ratelim...@lists.redbarn.org or http://lists.redbarn.org/mailman/listinfo/ratelimits instead of the dns-operations mailing list. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] How many kinds of DNS DoS attacks are we trying to stop ?
trust you have looked at other imperfect solutions such as rate limiting by (qname,qtype,IP address). In your estimation, how does that compare to trimming ANY and DNSSEC responses? Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] correction about RRL leakage
From: glen wiley glen.wi...@gmail.com This seems like a degenerate case to me...there is a threshold below which attacks are no longer meaningful. For most name servers I suspect that an attack is only interesting at some rate well above 10's of qps. The DNS RRL is less about defending a DNS server than the victims of the server. Only small or at most modest DoS attacks on a name server would be helped by dropping responses. One of the most effective family of DoS attacks against a name server is explicitly not addressed by the DNS RRL code. (There's no profit in enumerating attacks against DNS servers themselves or flogging their details here.) DNS RRL is mostly about mitigating DNS amplified reflection attacks in which an attacker bounces packets off DNS servers toward the real target and the DNS servers reflect or send many more bits toward the real target than they receive from the attacker. For example, a request for a DNSSEC validated A record for asdf.isc.org from a recursive resolver sends about 14 times as many bytes (~700) toward the supposed source than were in the original request (~50). As a name server operator not only am I not likely to see anything odd in an attack like that, I really don't have the time or inclination to care about volumes in that range. My DNS servers are certainly not what I'd call busy, but I'd probably not notice an extra 100 qps for days. However, a bad guy could send each of 1000 DNS servers 100 41-byte queries forged from 10.2.3.4 per second for a total of 32 Kbit/sec. Each of those requests would normally result in about 700 to more than 2000 bytes depending on the query. 10.2.3.4 would see 0.6 Gbps to 1.6 Gbit/sec. A discouraging fact is that rate limiting doesn't help if the bad guy uses a list of 100,000 or 1,000,000 servers and only 1 or 0.1 forged query/sec. The only hope is that by the time the bad guys get smart and ambitious enough to use millions of reflectors, BCP38 will be so common that the sending systems can be found and quenched. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] First experiments with DNS dampening to fight amplification attacks
From: paul vixie p...@redbarn.org first, since it does not take the query or response into account, all queries from a given source will share fate. this means your authority server will go completely silent on some recursive name server if it sends too many of any kind of query no matter how diverse those queries are. I would emphasize a different aspect of that issue. DNS Damping, like the firewall based schemes, allows a bad guy to silence a DNS server as far as a block of IP addresses is concerned. Silencing DNS has long been an important part of various security attacks. second, you go completely silent when in dampening mode. there are no slip responses by which an actual recursive name server might be able to get real answers by retrying with UDP or escalating to TCP during times that its IP address is being spoofed by an attacker. I think that is a feature of DNS Damping intended to answer the complaint about DNS RRL sending a constant data stream of attack packets. In my biased view that is a misfeature based on failing to apply the same DDoS scenario to DNS Damping. (See below.) third, you're giving each end-host address its own fate, so that a spoofed-source attacker could cause you to flood a distant network simply by iterating through that network's address space. There are references to dealing with blocks instead of individual addresses. Perhaps that is intended for a future version. your solution seems to be optimized for overly busy recursive servers who you want to deny excess service to, and does not deal at all with the case of spoofed-IP reflected amplified attacks. How so? i also note that you have misunderstood (and therefore mischaracterized) DNS RRL, according to this text from your web site: http://lutz.donnerhacke.de/eng/Blog/DNS-Dampening They can rate limit http://ss.vix.com/%7Evixie/isc-tn-2012-1.txt the queries per client. DNS Damping *is* rate limiting. It differs from DNS RRL only in details about counting and limiting rates. (One detail that I think is important but that Lutz Donnerhacke evidently does not is that RRL does not count *queries*; RRL counts *responses*.) Unfortunly this generates only a constant data stream of attack packets. DDoS works well with limited data rates per server, if you misuse enough servers. On the other hand the implementation required a lot of ressources. this text contains two factual errors: (1) that DNS RRL generates a constant stream of attack packets: we attenuate the attacks in two ways, first by dropping most (or at worse half) of the responses, second by responding with TC=1 packets that are no larger than the requests; and I think the complaint is that DNS RRL with slip 0 and the recommended responses-per-second 10 could send 10 DNS response/second to the victim. If the responses were DNSSEC ANY or NXDOMAIN results, they could be more than 1500 Bytes or 1.2 Kbits each. If you stop there, it sounds bad, because a bad guy could use 833 DNS servers using RRL to reflect 120 Mbit/sec to the victim. But don't stop there. Ask what happens if those 833 DNS servers use Damping instead of RRL. I think the same 120 Mbit/sec would be sent to the victim because no rate limiting including Damping should trigger on a busy server at much less than 10 requests/second. (2) that DNS RRL uses a lot of resources: we use about a megabyte of storage to keep unique state for 5 queries per second for five seconds, which is trivial. I also wonder about that claim that DNS RRL requires a lot of resources. From my superfical reading, appears to me that DNS Damping uses more resources than DNS RRL. What kind of resources are we talking about, memory or CPU cycles? How was resource utilization measured in the two implementations and what are the numbers? Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DNS RRL light?
- Rate limit clients to 100 qps. Drop for 5 mins if exceeded. - Rate limit client to 5 identical queries per second. Drop for 5 mins if exceeded. Any logical errors, or other errors, you see there? Also, any, simple to implement, enhancements you could add? The first rule is to do whatever works in your situation no matter what outsiders say. Are you counting identical queries including qtype as well as qname? Are you dropping identical queries or all queries? Counting all queries might not work on a server for a popular domain, because there can be a lot of legitimte queries from a single carrier grade NAT IP address. Even counting 5 identical queries or responses might cause problems for a sufficiently popular web site. Counting identical queries instead identical responses might let a bad guy reflect a stream of 1500+ Byte NXDOMAIN responses using a stream of queries for unique bogus domains. Blocking at 5 identical queries per second sounds reasonable to me, but blocking for 5 minutes sounds far too long, because it might unnessarily drop legitimate queries. A 5 minute window means that on average it will be closed 2.5 minutes after the attack stops. If your scheme can react to the first 5 identical queries in a second, why not block for only 10 or 15 seconds? Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DNS RRL light?
From: Mohamed Lrhazi ml...@georgetown.edu I am counting query_type+query_name, implemented as: set qhash [b64encode [md5 $q_type:$q_name]] I guess that's my state blob. Is that good? If possible, I'd use a bit hash function with lower CPU costs, although speed might matter. However, it sounds as this is all done as in shell scripts, and so it's not expected to handle many queries per second. Any good 128-bit hash will have no more collisions than a 128-bit cryptographic hash function and might have fewer. Contrary to ancient superstition, cryptographic hashes have no fewer or better distributed collisions than any other good hash function. Cryptographic hashes are only supposed to be hard to reverse and their collisions are supposed to be hard to predict. Those characteristics and their mundane collision characteristics are based only hope and a trivial number (compared to 2**128)of tests instead of mathematics like that behind other hash functions such as cyclic redundancy checks. If the IP address is handled elsewhere, any reasonable (not necessarily 'good') 32-bit hash such as `sum` or `cksum` should also have no collisions that matter, be a lot faster, and need fewer bits. The size of the state blob matters if you want to handle lots of queries/second and so need to store window size blobs; (10 seconds)*(100K blobs/second)=1M blobs. The key for each BIND9 RRL state blob consists of - 129 bits of IP address. Perhaps embedded IPv4 addresses would work, but this time I chose a separate IPv4/IPv6 bit. - qtype - simplistic 32-bit hash of qname to avoid a fixed size 256 byte buffer or worse, malloc - DNS class compressed to one bit for now - whether the response is an error, NXDOMAIN, or normal - whether TCP was used. Most of the BIND9 RRL blob consists of other stuff including links, counters, and timers. I am also dropping everything, during the drop window, as I did not want to keep the query info for too long, but since I will lower the window, it might be feasible. Dropping everything increases the likelihood of dropping legitimate requests, and so is another reason to make your window as short as possible. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DoS with amplification: yet another funny Unix script
From: Tony Finch d...@dotat.at I don't think diffuse is the right word - this kind of attack can be very intense. agreed, it's only diffused among qnames and qtypes. If you have a large domain signed with NSEC it's trivial for an attacker to enumerate the domain, and RRL will not treat this as an attack. Or of you are a large scale DNS hosting provider the attacker can get a list of domains you host from copies of TLD zones. Having got a list of names, the attacker can then reflect lots of traffic via your server which will be treated as OK by RRL. It would be easy to change the RRL patch to have yet another optional rate limit counting all non-error responses to an IP address block if they were the same. It would have some negative aspects: 1. Under that kind of attack, the TC=1 slipping is worse than useless, and so it would not trigger the TC=1 responses. 2. Targets of the reflection attack would get no DNS service at all unless they magically know to switch to TCP. 3. One can argue that this kind of defense belongs in a firewall that understands nothing about DNS except rate limiting based on source IP address and destination port 53. 4. It would double the memory spent on counting responses. The amount of memory required to count responses on very busy (10K or 100K qps) DNS servers has always been a concern. It is why the RRL patch saves a 4-byte hash of the qname instead using a 256 byte block (or worse, dynamically allocating space for each qname). However, it's only a factor of 2. I use the argument of #3 to respond to observations about the high costs of DNS/TCP and objections to TC=1 slipping. At sufficiently high rates, a DNS/TCP DoS attack looks like TCP SYN flooding. TCP SYN flooding is commonly handled without bothering the application and without allocating or timing-out a TCB in the either kernel or a firewall. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DNS ANY record queries - Reflection Attacks
From: Eric Osterweil eosterw...@verisign.com So, can I just make sure I understand the RRL idea? If, under non-attack circumstances, I get a traffic rate of `r' from a given subnet, but an amplification attack sends me `99*r' (causing a total traffic rate of `100*r'), then I should rate limit? So, my back of the envelope calculation says that I will reward the attack traffic over the non-attack traffic. That is, if I limit the response rate back down to `r', then I will drop 99/100 responses to reach that target. My legitimate client (subnet) has only about a 1/100 chance of getting each query answered here (all other response slots are given to my adversary)... That computation might be correct if DNS clients did not retransmit, if the BIND RRL idea involved only discarding responses, and if Paul and I proposed dropping 99% of all traffic for a CIDR block. We advocate none of that. We propose dropping only identical responses to a given CIDR block instead of all responses. The BIND RRL code has a notion of slip or responding while rate limiting with TC=1. It has a default slip rate of 2, or responding with TC=1 instead of dropping every other identical response. A DNS client that retransmits N times to a DNS server that answers 50% with TC=1 of the time will get an answer to 1-(0.5)^N of its queries. For N=4, it will get a TC=1 answer 94% of the time. I think rate limiting is kind of the wrong direction. Did I misunderstand some aspect? What do you think would be the right direction? Doing nothing is not acceptable. We think that rate limiting is only a work around for the failure of the responsible parties to implement BCP 38 or other effective mechanisms to stop the abuse the transmit on behalf of their users. In the distant future we hope it won't be needed. Also, when you say, ``shockingly effective,'' how can we measure effectiveness, in order to verify the approach? One way to measure the effectiveness of a defense is to compare the work the bad guy must do with the benefit to the bad guy. In this case, rate limiting at 10 identical repsonses and using the default {slip 2;} means that in common scenarios, the amplification is less than 1. The bad guy gets less result from a reflection DoS attack than a direct DoS attack. Under the circumstances, I think that is effective. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DoS with amplification: yet another funny Unix script
From: Klaus Darilion klaus.mailingli...@pernau.at On 10.09.2012 19:48, Paul Vixie wrote: please don't do, or promulgate, this. ddos filtering in order to do more good than harm has to be based on the attack's answer, not on its query. vernon schryver and i explain this in the technical note at http://www.redbarn.org/dns/ratelimits/. I fear that the technical note linked from that page fails to emphasize enough the drawbacks of firewall defenses against DNS reflection attacks. I was recently asked about a proposed security advisory about some consequences of using iptables to defend against DNS reflection attacks. My response was that it's not entirely the fault of the DNS server if a user's iptables facilitate denying DNS service. As I think that technical note says, any rate limiting has some danger of being exploited by bad guys, but simplistic firewall 'deny' rules are far too blunt and much too easily exploited. Any firewall rule that doesn't compute DNS responses about as good as a DNS server is simplisitic. If your firewall is smart enough to know that a stream of DNS requests will generate 1600 byte NXDOMAIN responses, why don't you turn off the DNS server and let the firewall do all of the DNS work? Is it correct that filtering based on responses is better only for NXDOMAIN responses or error responses. If the forged requests are requests for valid domain names, then analyzing the request should be fine too. Because each request implies its response, in theory there is no difference between filtering based on requests or responses. In practice it is impossible for anything except a DNS server to know the response that a request will generated. How can anything but the DNS server or its doppleganger know that a request will generate NXDOMAIN, SERVFAIL, REFUSED, or be dropped by an ACL in the DNS server? How can anything but the server know about variations in non-error responses due to views? There are few reason to forge requests that reflect as SERVFAIL or REFUSED, but you can get large amplifications with DNSSEC NXDOMAIN. If your defense works against non-error reflections attacks but not NDXDOMAINs, what will your adversaries do and how quickly? The party line that the BIND RRL only looks at responses is not entirely accurate. For obvious efficency reasons (and ease of coding), it looks at the RRsets or error code that would be used to generate the response instead of the final, on-the-wire response. Analyzing cooked instead of raw DNS responses is related to my main question about some of the suggested firewall schemes. How can you convert every incoming DNS packet to text and then run it through awk or grep and handle a non-trivial load? Of course it is easiest to track the states in the name server, but I do also see advantages in doing the filtering in an external node (upstream firewall, local iptables). For example I do not have to worry about the used name server software and I can use the same rules for bind, powerdns, nsd ... backends. I also suspect that filtering in kernel space may be faster. As a personal preference I try to separate functionality into dedicated nodes. That's an argument for all reasonable DNS server implementations having rate limiting knobs instead of a reason for bad and dangerous rate limiting in firewalls. As Paul said, the BIND RRL code is open. Why not remove all ACLs, views, and related mechanisms from your DNS server and put them in your IP firewall? The tuple mask(IP), imputed(NAME), errorstatus is used to select a state blob. In the amplification attacks on our authoritative servers we Thus, it may take some time until the attacker starts with domain1.com again. If I understand the Responder Behavior correct, this would mean that filtering is never triggered if a domain is not queried RESPONSES-PER-SECOND times per second. Or do I miss something here? I'm not sure I understand. If that points out that an attack that is too diffuse to be noticed by the BIND RRL code might be noticed by a firewall rule, then I agree. I'd also say that can be seen as a feature instead of a defect, because during less diffuse attacks, legitimate requests from the forged CIDR block will still be answered. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DoS with amplification: yet another funny Unix script
From: =?ISO-8859-1?Q?Colm_MacC=E1rthaigh?= c...@stdlib.net Any firewall rule that doesn't compute DNS responses about as good as a DNS server is simplisitic. With the greatest of respect; that thinking is itself simplistic. Where I work we concentrate on writing very good firewalls. Sometimes these rules even have to parse DNS, just as the DNS server must ... That just as the DCC server must is false. For example, I doubt that those firewalls do enough DNS computing to recognize and limit a stream of responses generated from a single wildcard before the responses have been transmitted by the DNS server. They probably doesn't even recognize pernicious but simple NXDOMAIN cases. They might but probably don't notice that a stream of responses are approximately identical referrals from authoritative servers or approximately identical recursion from recursive servers. I think DNS rate limiting must do all of that while not slowing other high volume traffic. An in-the-path firewall actually has access to more data than the DNS server alone does. For example, it can build up a simple profile of expectation values for IP TTLs on a per-source network basis. That is also overstated. In practice DNS servers don't do such things, but they could. (Yes, you can get UDP/IP headers in a modern BSD UNIX daemon.) I doubt that the computing costs of tracking IP TTLs would be worthwhile for a DNS server with high legitimate load. I wonder if administrative costs such as dealing with the false positives due to route flapping or re-homing would be worthwhile even in a firewall. Remember the best of all firewalls that is advetised on http://www.ranum.com/security/computer_security/papers/a1-firewall/ It can use all IP data for that profile; DNS, HTTP, whatever it's seen. Those expectation values can then be used to detect and reject spoofed packets, in combination with other statistical scores. That's just one simple example - there are many more. Firewalls have good and valuable uses in lower layer defenses. However, firewalls are usually weak crutches for applications. They are very popular for quick plugs in application holes, because badly designed and written applications are so popular and it's a lot easier to kludge something into a firewall than fix the typical lame application code implementing a worse than stupid de facto standard protocol. Even good protocols have weaknesses. For example, every protocol and especially those using UDP must have basic features including: - optional authentication authorization - exponential or steeper backoff for retries - rate limiting on requests from evil as well as innocently broken clients The original DNS lacked all of those. The other big reason is pragmatism; unix daemons using recv() are extremely limited in the rate at which they can process packets. far far higher throughput is possible via other techniques that involve handling batches of packets at much smaller timescales. A nice benefit of the approach is that it frees higher-level development teams from having to worry about low-level mitigation, and that the work is re-usable across many products. During real attacks, if a packet makes it to the dns server, the game is already lost. I disagree with most of that. Since it is about general philosophies and what is theoretically possible instead of operational issues or even DNS theories, I'll resist the impulse to pick it apart. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DNS ANY record queries - Reflection Attacks
From: Eric Osterweil eosterw...@verisign.com That computation might be correct if DNS clients did not retransmit, if the BIND RRL idea involved only discarding responses, and if Paul and I proposed dropping 99% of all traffic for a CIDR block. We advocate none of that. Hmm.. I may still be missing some nuances, what are the specifics? Instead of my writing a new complete description or even worse for other readers, copying-and-pasting the existing documentation, please see the recently mentioned http://www.redbarn.org/dns/ratelimits That page has a link to a technical note talking about the problem, a link to code for serious specifics, and even a link to draft changes to the BIND9 ARM. So, I don't understand something... If you see a lot of identical responses from an authority, could that not be because it is an authority for those responses? How do you distinguish a netblock with multiple resolvers, or anycast resolvers? The BIND RRL code is part of the resolver. It does not see a lot of identical responses from an authority except when it is the authority. Perhaps more directly, are you dropping responses from legitimate clients and how do you feel about them being collateral damage? Paul Vixie and I are not advocating DNS rate limiting in firewalls. We're talking about rate limiting in the hosts at the ends of the intertubes. So, every identical response either gets dropped or gets its TC bit set? No, every *excessive* identical response is either not sent (dropped) or a tiny TC=1 response is sent instead. A DNS client that retransmits N times to a DNS server that answers 50% with TC=1 of the time will get an answer to 1-(0.5)^N of its queries. For N=4, it will get a TC=1 answer 94% of the time. Wait, I'm very confused... The above sounds like you respond to 94% of the reflector attack queries (which furthers the attack). No, I was pointing out that P(R1R2R3R4)=P(R1)*P(R2)*P(R3)*P(R4) Given a uniform drop probability of 50%, the probability that all 4 responses to an initial request and its 3 retransmissions will dropped is 6%. (Or should that be N=5?--I always seem to be off by 1. In which case 97% of requests would be answered.) Well, if doing something hurts the legitimate clients more than doing nothing, I think you need to be upfront about that. I think that's worse than doing nothing. That's like opposing mechanical spam filtering by pointing to mechanical false positives while ignoring the higher false positive rate of the otherwise inevitable purely manual filtering on subjects and senders. If you do nothing, then legitimate clients will be denied all service by the firewall rules advocated here or by IP bandwidth rate limits at the source (DNS servers) and the DoS targets. Remember why it's called a DoS. You are saying that you would rather try to receive 1000 1500 Byte bogus DNS responses per second along with all your legitimate DNS responses that don't get dropped from router queues by that flood instead of 10 bogus responses and useful responses to 94% of your requests. OK, but you've also almost certainly eliminated the legitimate client's ability to query you for responses. That is simply false. When Paul Vixie wrote that the BIND RRL code is effective, he wasn't talking about theory or small scale tests. It has been in use on some major DNS servers for months. If there were enough collateral damage to talk about, someone would have complained. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DNS ANY record queries - Reflection Attacks
From: Eric Osterweil eosterw...@verisign.com Fair enough, except I'm pretty sure some of the deployment being talked about (even in this thread) is at the authority (not the resolver)... Paul Vixie and I are not advocating DNS rate limiting in firewalls. We're talking about rate limiting in the hosts at the ends of the intertubes. Again, this thread started somewhere else. Clearly, I agree that people should be able to manage their own user experiences. ;) By definition, all DNS servers and clients run on hosts. When a router or bridge does DNS stuff, it is a host and subject to the Host Requirements RFCs. I and I think some others have been talking about filtering DNS requests and/or responses in the following hosts and/or firewalls close to those aforesaid hosts: - DNS servers, either authority servers or resolvers - putative DNS clients that are targets of DNS reflection DoS attacks. The money-back warranty on the BIND RRL patch only covers its installation on authority DNS servers, although I have received positive reports from resolver operators. The current version includes additional features suggested by operators of combined authorities/resolvers. 1 - If you uniformly drop 50% of a 100x amplification attack, you are still reflecting 50x amplification, right? That is wrong for the BIND RRL patch. With default parameters, the BIND RRL drops 50% of responses and substitutes a small TC=1 response for the other 50%. That gives an amplification for the responses it sends of = 1.0 responses sent and an overall amplification = 0.5. (It currently forgets about ENDS in the TC=1 responses giving a default amplification of 0.5. An influential commentator calls that a bug.) 2 - If you wait for (say) 4 responses, your stub (the client driving the upstream resolver) has almost certainly timed out, and the DDoS has succeeded, if I'm not mistaken, right? I think that is mistaken. Consider the implications of the default values of attempts and timeout keywords in /etc/resolv.conf and of _res.retry and _res.retrans the libc interface to the de facto standard stub. Then it should be easy enough for someone to explain the above, no? Having deployed something does not mean that it was effective, and blocking traffic does not tell me how much legit traffic and how much attack traffic was blocked. I don't see why this is so hard, I just want to understand the assertion. Consider where the BIND RRL patch has been installed and then ask yourself why *you* have not noticed any collateral response losses. See https://lists.dns-oarc.net/pipermail/dns-operations/2012-June/008453.html Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Research Project: Identifying DNSSEC Validators
From: =?ISO-8859-1?Q?Matth=E4us_Wander?= matthaeus.wan...@uni-due.de I assume I'm odd, because I'm not eagar to put the invisible HREF anchor on my web pages because of the extra DNS transactions imposed on users. I also have vague worries I can't articulate about privacy concerns. =20 My answer to putting a simple IMG beacon on my web pages would be a flat never. There are too many technical and legal issues. For example, what about privacy issues with the referer string? =20 Can't argue with that. If privacy is an issue, you won't become friends with foreign HTTP resources. I don't understand that. Whether an HTTP server is foreign or domestic (for any value of domestic) does not by itself determine its trustworthiness. I start by assuming any HTTP server is untrustworthy, but that doesn't imply that I should involve third parties. The privacy issues I meant involve the third parties counting DNSSEC aware resolvers. The commercial hit counters also claim to be trustworthy, even as they sell their measurements. I assume that none of you guys would do something like correlating referer strings, your results, and WHOIS or other e-appended values to send email to web masters offering to sell better DNS resolver software. I also assume that if a financial institution put your beacons on their TLS web pages, none would try to 'leverage' the resulting referer, weak DNS resolver, and IP address data. And so forth and so on including other attacks I can't imagine. However, a security policy based on assumed good intentions is incompetent. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Side effects of enabling DNSSEC?
From: Mohamed Lrhazi ml...@georgetown.edu I am learning DNSSEC and was wondering if there was any side effects to enabling DNSSEC on a domain, if there were mistakes in the configuration? In other words, if I were to enable DNSSEC on a zone, and miss something, could I effect anything other than DNSSEC validation itself? and if I did effect it, how bad would that be? and also, how would you go about testing that everything is working fine once enabled? If the signatures don't work, then resolvers that pay attention to DNSSEC will answer requests for your DNS records with SERVFAIL. I guess I should ask the same question about side effects when there are no configuration mistakes at all :) Should I expect anything to break because now DNSSEC is enabled and working? More stuff always means more stuff that can and so will break. I think that there are better questions: 1. Will you ever enable DNSSEC on your domains? 2. If so, should you do it now or later? #1 is for you to answer. If your answer for #1 is yes, then the answer for #2 is that you should join the DNSSEC party yesterday or today at the latest. Because current versions of BIND pay attention to DNSSEC by default, ever larger fractions of the Internet and eventually most of it (well, outside the jurisdictions of 'authoritative regimes') will penalize DNSSEC errors with SERVFAIL. I suspect that today only a large minority the Internet does that, and so now is the time to learn from the inevitable mistakes. Today you might need to use `dig +dnssec` or `dig +adflag` to see the effects of DNSSEC. Tomorrow you will need to use `dig +cdflag` to not see them. For example, I get a long delay and then SERVFAIL for `dig www.dnssec-failed.org` from a resolver with the BIND9.9 default for dnssec-validation. A real life instead of artificial example also found on http://www.dnssec.comcast.net/ is the quick SERVFAIL that I get from `dig usbountyhunters.com` Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
[dns-operations] DNSSEC, IPv6 glue, multiple DNS servers, and eating your own dog food
An obvious filter for prospective registrars occurred to me at this, nearly the end of my week long efforts to get my trivial domains signed. A registrar that does not have DS records for its main domain names might lack experience dealing with DNSSEC registrations. A registrar whose main domains lack records for any NS names might lack real world information about IPv6 glue. A registrar or reseller that does not have have a WHOIS record with a minimal set of servers or at least NS records might lack empathy for those of us who think such things are a good idea. You don't need to ask people at the registrar. `dig example.com ds`, `dig example.com `, and `dig example.com ns` can give more authoritative answers than anything people might say. Its funny but not amusing that those commands give better results for the unreal example.com than for any of the registrars that I recall being mentioned here recently. This should be particularly embarrassing for one of them. I tried several other registrars on http://www.dotandco.net/ressources/icann_registrars/details/position.en and found *none* that could pass that trivial filter. Talk about a race to the bottom! Vernon Schryverv...@rhyolite.com P.S. eating your own dog food is not an insult but the old programmer's motto about using the stuff that you would foist on others. P.S. The imperfections in ARIN's reverse DNS web page are consistent with the lack of DS RRs for arin.net. ICANN passes. ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] thoughts on DNSSEC
It seems to me that registrants are in a position to make determinations about the extent of a regisrtar's actual support of DNSSEC on the basis of their interface support and the claims of their support staff. How does a prospective customer check a registrar's interface without doing something approaching reality like registering a throw-away name? The costs in time and hassles of that are a barrier. Worse, not only could send mail be better in theory than web forms, you can't tell whether web forms are sugar coatings on dung worse than typical send mail without a lot more than an initial registration and twiddling. For example, from an initial registration I wouldn't have known about GoDaddy's understanding of permission to try to bill credit numbers they happen to have for wonderful new services like alert mail boxes (nothing to do with SMTP, IMAP, or POP). For another example, Opensrs/Tucows claims to support IPv6, but when I last tried an IPv6 address in the store front web forms somehow provided to my reseller by Opensrs/Tucows, they choked. You win if you bet that the Opensrs answer was send mail with your glue to support. I blame my reseller for not simply diving in and fixing the web form parsing, but not more than I value past personal attention and advocacy inside the registrar. The ICANN pages are surely a good place to start, but only to start. I think that overstates the value of the ICANN page. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] question for DNS being attacked
From: Michael Graff mgr...@isc.org It may also make Kaminsky style attacks easier if an attacker can blind an auth server from handing out responses. If the counter values are real from the RFC style paper, every other response becomes a truncated reply in a flood situation. This will extend the attack window by the time it takes to establish a TCP connection and query, or to the time it takes to retransmit the query plus TCP handshake if the blinding is successful. This assumes the second query works. But in reality it has the same chance as the first. Assuming about 100ms for TCP handshake and two seconds for timeout and retry followed by the TCP handshake, This means the window for potential false responses moves to about 1100ms on average. If a UDP reply would normally make it to the server in 100 ms, this opens a window 11 times as wide. That conclusion does not hold, because it does not define the narrow window alternative. 11 times as wide as what? Causing a DNS server to stop answering your target is a security issue, but the alternative to rate limiting based on (IP,qname,type,class) is not infinite bandwidth. If DNS server software does not rate limit its own answers, then its answers will be rate limited by bottlenecks in the path between the server and the client. This is true whether the bottlenecks are in firewalls or routers. To prevent an authoritative server from answering your requests, I need only trigger the rate limiting between it and you. If the rate limits are based only on bandwidth, then I will cause it to try to send more than that limit. Before botnets and DNSSEC, that often would have been difficult. Today I can send as many requests/sec as I want and I can use DNSSEC to make the responses at least 20 and often 50 times larger than the requests. Consider choices for IP bandwidth limits. Because of the 20X-50X amplification of DNSSEC, aren't ipfw/iptables limits likely to be correspond to about the values that you would choose for (IP,qname,type) limits? What is the difference in a security attack whether an auth server is silenced by random DNSSEC requests that overrun its output bandwidth limit or its (IP,qname,type) limit? I see these differences: - Depending on the security attack, the attacker might not know the qname. In that case, the attacker must fall back to exploiting bandwidth limits. - During the attack, the server might still answer other queries if it uses (IP,qname,type) limits. Another way of stating this difference is that relying on bandwidth limits to stop participation in DoS attacks makes it easier for an attacker to completely silence a DNS server in other kinds of attacks. - (IP,qname,type) rate limiting makes it less likely that your DNS server will be used as part of a DoS attack. (IP,qname,type) rate limiting is not a silver bullet and IP bandwidth rate limiting using fair or any other queue discipline is not magic pixie dust. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] question for DNS being attacked
From: Michael Graff mgr...@isc.org That conclusion does not hold, because it does not define the narrow window alternative. 11 times as wide as what? With a slip factor of 2, every other packet will be dropped, and the = other packets returned will have the truncated bit set. If this is = incorrect, please explain what it does do. This translates to a normal truncated response 50% of the time, and a = timeout the other 50%. That is true only if you modify the patch to accept a rate limit threshold 0. Otherwise the truncated responses and timeouts happen only during attacks involving flooding, including what I assume are meant by spoofing flood attacks. I don't think running with a rate limit threshold 0 makes any sense, but I've often been called lacking in imagination. If I should have understood normal as normal during a spoofing flood attack, then we can talk about what happens with IP instead of (IP,qname,type) rate limiting and the thresholds for particular instances of the universal IP rate limiting. Ignoring the penalty BIND 9 and other servers = are likely to assign to this misbehaving server, the timeout keeps the = Waiting for a response window open much, much longer. Again, longer than what? The alternative is *not* an absense of rate limiting and so an absense of windows during which the target is prevented from receiving responses from the real authoritative server. As I said before, whether one uses (IP,qname,type), IP bandwidth rate limiting or both, the target can be prevented from getting response from authoritative servers. The only things that vary are the contents of the packets used by the attacker to blind the target, perhaps (but not necessarily) the bandwidth used by the blinding packets, and whether the thresholds for the rate limiting were consciously chosen to deal with any kind of DNS issue. It's a mere implementation detail of the attack whether the window is held open by the authoritative server's refusal to answer or by configured or de facto bandwidth rate limits in the server or the path from the server to the target. A separate aspect of this supposedly much, much longer window is that it seems to assume that after the client has received a truncated or TC=1 response and is going through the DNS/TCP dance, it will still accept forged, evil DNS/UDP responses from the attacker. Is that true of common resolving servers and resolver libraries? It's not how I'd write a resolver. Instead I'd discard all of the state needed to accept apparently stale UDP responses before the TCP SYN is sent. Waiting for a response window open much, much longer. This timeout is = largely server-dependent, and some may wait multiple seconds. This = leaves the window open for a spoofing flood attack to sneak in. Again, longer than what? While I commend you and Paul on the RLL work you've made, I think it is = improper to not mention this in the documents you write. It may be not = a big deal to the administrator of the zone, but it is up to them to = decide that. Some may prefer to be a flooding source rather than make = their zone more prone to spoofing, even if the actual odds are low. The = biggest problem here is that the zone publisher's goals of not being = spoofable are entirely dependent on the resolver asking the questions, = without DNSSEC in the mix. The notion that there is an alternative to rate limits is wrong. There are rate limits in the path from the server and to the client whether anyone has consciously configured them below the limits imposed by service bit rates. The lowest limit might be in firewalls near the server or the client or in intermediate router queues, but it will exist and can often (almost always?) be utilized by an attacker using the 20x-50X amplification of DNSSEC. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] ok, DNS RRL (rate limits) are officially, seriously, cool
From: Klaus Darilion klaus.mailingli...@pernau.at Nice. But I wonder why there is a drop-down of outgoing packets during an amplification attack. I would expect that outgoing traffic is constant. Maybe, in this case also legitimate queries are blocked (false positive). Why would false positives happen only while there are lots of true positives? This rate limiting scheme is not an automatic IP address or domain name ACL. The only likely false positives are legitimte requests both for the same records as the attack requests and from the same IP as the forged requests. If the reduction indicates false positives, then the bad guys are forging requests that are from real clients and for the same names, but not by themselves enough to the reach rate limit. So I think the reduction could be false positives only if the attack involves a lot of differing client IP addresses and some very popular names. Note also that the graphs don't say whether a reduction in outgoing packets happens during an attack without the rate limiting. The reasonable guess is that somewhere in the path from the real and attacking clients up to and including the server there are bottlenecks that let the attack hurt legitimate traffic, but we don't know where. Maybe the attack blocks legitimate requests in router or firewalls between the server and legitimate clients. From: Phil Regnauld regna...@nsrc.org } That's assuming all other clients are behaving properly in the } first place, could be a non negligible number of malware generating } this background noise. Their existence might be revealed by rate } limitation. Because this rate limiting scheme is not an automatic IP address or name ACL, I don't understand how that might happen. Why would bad guys be continuing forging about 1 qps for the same clients and name as during the real attack? } But yes, it's worth digging. Agreed. However, the obvious test of checking for a reduction in legitimate responses during an attack would be hard (how could you tell?) and unsavory. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Why would an MTA issue an ANY query instead of an MX query?
From: Florian Weimer f...@deneb.enyo.de Emergency patches against ANY to last for a day or two for lack of other available tools can make good sense--for a day or so. But spending any long term effort on ANY queries in this context is the same thinking that brought us SPF as the final ultimate solution to the spam problem (FUSSP), because as we all knew, spam requires forged senders. But unlike spam, these attacks require spoofed source addresses. Was I really that unclear? Of course forged IP source addresses are a critical part of DNS reflection DoS attacks, just as bulk is a critical part of spam. My point is that it is necessary to pay attention to the necessary aspects of the problem and deal with those instead of trivial efforts against the current wrapping paper. From the history of obvious bogus spam FUSSPs such as the many variations of email authentication and the prove mail sender is a human unsolicited bulk email (spam) sent to uninvolved third parties, I predict that the next solution to DNS reflection attacks after the current disable AUTHORITY and ADDITIONAL sections and disable ANY will be disable DNSSEC. Solutions analogous to know your customer before allowing outgoing bulk connections to TCP port 25 such as disable or restrict open recursive DNS servers to known users or even install response rate limiting DNS software (not to mention BCP 38) are resisted as too hard. The saving grace is that the monetary rewards for allowing DNS reflection attacks aren't as large as those for allowing unsolicited bulk email. Perhaps it's time to admit defeat, call our legislators, and suggest that they mandate source address validation by service providers. Speaking of easier non-solutions that would not only not solve the problem but create worse problems ... On the other hand, if service providers were liable for damages caused by forged IP source addresses (or forged SMTP envelopes) ... Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] dns response rate limiting (DNS RRL) patch available for testing
From: Ken A k...@pacific.net To: dns-operati...@mail.dns-oarc.net On a authoritative + recursive server, instead of a separate view, we use: acl trusted { x.x.x.x/z; }; allow-recursion { trusted; }; Is there any way to apply this patch so that it does not affect a specific acl, such as trusted addresses? Or, is it recommended/required that we configure separate views to use this? Separate views are required to apply rate limiting to some but not all DNS clients, unless you are of the school that holds authoritative+recursive servers are always utterly wrong. In that case separate servers are required. Would it be easy to transform your configuration file to use views via the include directive? My named.conf files look something like view insiders { match-clients { goodnets; }; recursion yes; include privatezones; include publiczones; response-policy { ... }; }; view outsiders { match-clients { any; }; recursion no; include publiczones; rate-limit { ... }; }; Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Why would an MTA issue an ANY query instead of an MX query?
From: Tony Finch d...@dotat.at Yes, how is BCP 38 deployment going? Someone on NANOG recently mentioned http://spoofer.csail.mit.edu/ http://rbeverly.net/research/papers/spoofer-imc09.html and the last slides in http://rbeverly.net/research/papers/spoofer-imc09-presentation.pdf suggest that relying on BCP 38 deployment is unsound. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] DDoS botnet behaviour
From: Tony Finch d...@dotat.at To: Vernon Schryver v...@rhyolite.com cc: dns-operati...@mail.dns-oarc.net The reason I'm basing my work on a Bloom filter is to avoid any per-client scaling costs. There's a fixed per-packet overhead, a fixed memory cost (which should be scaled with the server's overall load), and a fairly cheap periodic cleaning task. No dynamic memory allocation. How many hash functions are you using, what are they, and how do you know that they are sufficiently independent to give a tolerable false positive rate without using as much RAM as a single classic hash table? I hate dynamic memory allocation, and so it doesn't happen in my code after it reaches steady state. The operator can use an optional parameter to set initial sizes to minimize the cold start problem. What is your estimate for the memory required for a simple hash based scheme? Memory size even on the busiest servers was a primary design goal for my code and I was told to handle 100K qps. Since another goal is to stop dropping responses the instant the forged requests cease to avoid being a DNS Dos for the target, the rate must be computed with only very recent experience (which eliminates the ancient BSD alpha decay function). That means that one doesn't need more than a (very few seconds)*(100K qps)*(memory per response). I don't think much of timers, and so currently my code has none. My rate counters are what might be called self-clocking. I think there will eventually be a timer for logging, which should save a lot of cycles when logging is used and cost nothing when logging is not used. Your operational criticisms of the probabilistic approach are quite correct. It may also turn out to cost too much to get an acceptably low false positive rate. Do you believe that single set of Bloom filter parameters can serve an interesting class of installations without using more RAM than a classic hash function? To understate it--I don't. How does one learn the false positive rate on an operational system without running a deterministic rate limit system in parallel? If you can make the false positive monitor work well enough to be used everwhere it must be used including at 100K qps, what's the point of the Bloom filter? But, it might be worth putting a smallish Bloom filter in front of an accurate traffic accounting system, so that the server only needs to spend resources tracking the heaviest users, along the lines described in http://pages.cs.wisc.edu/~estan/publications/elephantsandmice.pdf Bloom filters are a neat idea for the right jobs. I also think that the 2 or 3 (not to mention more) hash lookups in a Bloom filter would use far more CPU cycles per query than needed for an accurate accounting system. When not logging and after reaching steady state, my code only hashes the request, changes pointers at most 3 linked lists, increments a counter, and tells the caller in client.c or query.c to continue or drop. Isn't that about what each of what 2 or more Bloom filter would hashes do? Bloom filters combine several poor hash functions with a single, extremely small (in context) bit array to give answers that are imperfect but amazingly good. That's cool, but not a mandate for replacing all other hashes, trees, tries, etc. Vernon Schryverv...@rhyolite.com ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
Re: [dns-operations] Why would an MTA issue an ANY query instead of an MX query?
From: Tony Finch d...@dotat.at I think it's wrong to focus on ANY queries: restricting them just encourages the attackers to move on to another query type. For a domain with DNSSEC you get almost as much data in return to an MX query - 2KB vs 1.5KB for cam.ac.uk. Today I see 2232 byte responses for another type from the authoritative servers for another domain often discussed in this context. That obvious type is not TXT, SPF, MX, or anything else that might be deleted, deprecated, shrunk, compressed, moved to an apex, or whatever. ANY queries might be of little use to computers, but I find them useful while chasing DNS problems. Emergency patches against ANY to last for a day or two for lack of other available tools can make good sense--for a day or so. But spending any long term effort on ANY queries in this context is the same thinking that brought us SPF as the final ultimate solution to the spam problem (FUSSP), because as we all knew, spam requires forged senders. That analogy goes farther than one might realize, because some of the ANY solutions I've heard include analogs of the amazingly uninformed and wrong headed SPF re-invention of SMTP source routes. Vernon Schryverv...@rhyolite.com P.S. I know the current line is that SPF is not and never was a FUSSP; that doesn't change what was said at the time. I also know that DKIM has some real operational value, despite the fact that plenty of unsolicited, objectionable bulk email advertising is delivered with valid DKIM signatures. ___ dns-operations mailing list dns-operations@lists.dns-oarc.net https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs