Re: Tata Scenic routing in LAX area?
Marcus, From route-views output, it looks like AS9498/airtel is probably leaking your route between two of its upstreams (AS6453/Tata and AS4637/Telstra) overseas, funneling some of your traffic through their router. route-views>sh ip bgp 23.92.178.22 | i 9498 3356 6453 9498 4637 29791 1403 6453 9498 4637 29791 3549 3356 6453 9498 4637 29791 19214 3257 6453 9498 4637 29791 1403 6453 9498 4637 29791 286 6453 9498 4637 29791 53364 3257 6453 9498 4637 29791 3257 6453 9498 4637 29791 1239 6453 9498 4637 29791 2497 6453 9498 4637 29791 57866 6453 9498 4637 29791 7660 2516 6453 9498 4637 29791 701 6453 9498 4637 29791 3561 209 6453 9498 4637 29791 You might try halting advertisements to your AS4637/Telstra peer while you contact AS9498. -John On 11/15/2018 10:43 AM, Marcus Josephson wrote: Anyone else seeing an odd Scenic routing in the LAX/SJE area for tata. traceroute to 23.92.178.22 (23.92.178.22), 30 hops max, 52 byte packets 1 if-ae-13-2.tcore2.lvw-los-angeles.as6453.net (64.86.252.34) 180.698 ms 180.610 ms 181.712 ms MPLS Label=344269 CoS=0 TTL=1 S=1 2 if-ae-7-2.tcore2.svw-singapore.as6453.net (180.87.15.25) 189.327 ms if-ae-7-2.tcore2.svw-singapore.as6453.net (64.86.252.37) 176.800 ms if-ae-7-2.tcore2.svw-singapore.as6453.net (64.86.252.39) 174.631 ms MPLS Label=609315 CoS=0 TTL=1 S=1 3 if-ae-20-2.tcore1.svq-singapore.as6453.net (180.87.96.21) 174.287 ms 173.370 ms 173.804 ms 4 120.29.215.202 (120.29.215.202) 179.104 ms 179.367 ms 179.324 ms 5 182.79.152.247 (182.79.152.247) 180.164 ms 182.79.152.253 (182.79.152.253) 184.816 ms 182.79.152.247 (182.79.152.247) 250.928 ms 6 unknown.telstraglobal.net (202.127.73.101) [AS 4637] 173.974 ms 173.986 ms 173.484 ms 7 i-93.sgpl-core02.telstraglobal.net (202.84.224.189) [AS 4637] 175.094 ms 175.699 ms 174.343 ms 8 i-10850.eqnx-core02.telstraglobal.net (202.84.140.46) [AS 4637] 280.686 ms 288.703 ms 280.836 ms 9 i-92.eqnx03.telstraglobal.net (202.84.247.17) [AS 4637] 278.021 ms 276.637 ms 302.249 ms 10 equinix-ix.sjc1.us.voxel.net (206.223.116.4) 174.139 ms 174.163 ms 174.067 ms Marcus Josephson IP Operations mjoseph...@inap.com /This message is intended for the use of the intended recipient(s) and may contain confidential and privileged information. / /Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact//the sender by reply email and destroy all copies of the original message./
Re: Amazon network engineering contact? re: DDoS traffic
Zach, Yes, RTBH is used to distribute the null-routes that I mentioned. Unfortunately, even brief saturation events lasting just 5-10 seconds (a typical amount of time to detect the loss, issue the null-route, and see the traffic start to fall off as it is distributed upstream) can cause real damage to those customers who are sensitive to latency and packet loss. So while null-routes limit the duration of the impact, they can't eliminate it entirely. And, of course, the actual target of the attack -- the now-null-routed IP address -- becomes unreachable, which was presumably the goal of the attacker. -John On 11/8/2018 12:54 PM, Zach Puls wrote: No idea about an Amazon abuse contact, but do you have RTBH communities enabled with your upstream provider(s)? As a general practice, when you detect a (D)DoS attack in progress, it would help to automatically advertise that prefix to your upstream(s) with the black-hole community. This would at least help mitigate the effects of the attacks when they do occur, even if they come from a different source than AWS. Thanks, Zach Puls Network Engineer | MEF-CECP KsFiberNet -Original Message- From: NANOG On Behalf Of John Weekes Sent: Thursday, November 08, 2018 14:44 To: nanog@nanog.org Subject: Amazon network engineering contact? re: DDoS traffic We've been seeing significant attack activity from Amazon over the last two months, involving apparently compromised instances that commonly send 1-10G of traffic per source and together generate Nx10G of total traffic. Even when our overall upstream capacity exceeds an attack's overall size, the nature of load-balancing over multiple 10G upstream links means that an individual link can be saturated by multiple large flows, forcing our systems to null-route the target to limit impact. We've sent an abuse notification about every traffic source to Amazon, and specific sources seem to stop their involvement over time (suggesting that abuse teams are following up on them), but there is an endless parade of new attackers, and each source participates in many damaging attacks before it is shut down. Is there anyone at Amazon who can help with an engineering solution in terms of programmatically detecting and rate-limiting attack traffic sources, to our networks or overall? Or applying the kludge of a rate-limit for all Amazon traffic to our networks? Or working with us on some other option? At least one other large cloud provider has an automatic rate-limiting system in place that is effective in reducing the damage from repeat high-volume attacks. Emails to the Amazon NOC, peering contacts (since that would be another possible solution), and abuse department have not connected me with anyone. Thanks, John
Amazon network engineering contact? re: DDoS traffic
We've been seeing significant attack activity from Amazon over the last two months, involving apparently compromised instances that commonly send 1-10G of traffic per source and together generate Nx10G of total traffic. Even when our overall upstream capacity exceeds an attack's overall size, the nature of load-balancing over multiple 10G upstream links means that an individual link can be saturated by multiple large flows, forcing our systems to null-route the target to limit impact. We've sent an abuse notification about every traffic source to Amazon, and specific sources seem to stop their involvement over time (suggesting that abuse teams are following up on them), but there is an endless parade of new attackers, and each source participates in many damaging attacks before it is shut down. Is there anyone at Amazon who can help with an engineering solution in terms of programmatically detecting and rate-limiting attack traffic sources, to our networks or overall? Or applying the kludge of a rate-limit for all Amazon traffic to our networks? Or working with us on some other option? At least one other large cloud provider has an automatic rate-limiting system in place that is effective in reducing the damage from repeat high-volume attacks. Emails to the Amazon NOC, peering contacts (since that would be another possible solution), and abuse department have not connected me with anyone. Thanks, John
Re: Spitballing IoT Security
On 10/29/2016 9:43 PM, Eric S. Raymond wrote: I in turn have to call BS on this. If it were really that easy, we'd be inundated by Mirais -- we'd have several attacks a*day*. Some of us are seeing many significant attacks a day. That's because botnets are frequently used to hit game servers and game players. In fact, the Mirai-targeted devices were not newly-seen; easily-exploited devices like older DVRs have been observed for years in attacks on game servers. The main difference in the recent botnet attacks (mostly, 2016) is that they have been larger and more frequent, likely because of incremental improvements to scanners (including in time-to-exploitation, which is important to building the botnet because these devices are so frequently rebooted) and payloads (to better block further exploitation by competitors). If you run a honeypot and take a look at what happens to one of these devices over time, you'll see an interesting tug-of-war between many different actors that are compromising them and running their own binaries. Reflection attacks are still common, as well, of course. Previously, those were the ones that created the largest flows. But, the higher-amplification-factor reflection attacks can be mostly mitigated upstream with basic ACLs (as long as the upstream is willing to help, and has the internal capacity to do it; many NSPs do not). It is not uncommon to see a botnet attack at the same time as a reflection attack. -John
Re: Death of the Internet, Film at 11
On 10/24/2016 9:37 PM, b...@theworld.com wrote: As I've suggested before how much would you attribute this to a lack of English skills by recipients? I do not think that is a significant factor. Here are some points along those lines: - ab...@cnc-noc.net times out. It's not a matter of whether they know English; they just don't accept the email. - Some Hong Kong ISPs /do/ respond and ask questions. In English. (As does a sampling of other foreign ISPs around the world, including those in Japan, Europe, Russia, etc. -- but mainland China is consistently silent.) - The major Chinese players (including China Mobile, China Telecom, and China Unicom) are some of the largest companies in the world, with just China Mobile having 241,550 employees, according to their 2014 annual report. It is unlikely that they don't have internal translation capabilities. I also have no doubt that they have a large NOC, and they could have a large abuse team (but perhaps choose not to). Large teams are more likely to have some bilingual members, and English is a very common second language. - These large Chinese companies are global companies with PoPs inside the U.S, and peering with U.S. providers. They sell services to, and interact with, companies around the world, including in English. - I have had others tell me that engineers at these Chinese providers contact them for peering upgrades in English -- but that they ignore abuse concerns communicated over the same channels. - Knowing English is not necessary to read tcpdump output, recognize attack traffic, and check IP addresses. Recipients don't have to respond back, so that's mostly what they need. - It's not hard to use online translation services. - It's not hard to respond back and say "Use Mandarin" (or the equivalent, in their preferred language). - I tried sending emails to Russian providers in Russian for a time. I received quite a few responses back along the lines of "please just use English." This has made me think twice about trying to pre-translate. Are they all sent in English? Currently, mine are. Just curious but one wonders what most here would do with an abuse complaint sent to them in Chinese? If I were to receive one in Chinese, I would personally paste it into Google Translate. That is what I do with Japanese complaints/responses, which are the main ones I see that aren't in English. Most others ISPs seem to use straight English, or both English and another language. -John
Re: Death of the Internet, Film at 11
On 10/23/2016 4:19 PM, Ronald F. Guilmette wrote: ... I've recorded about 2.4 million IP addresses involved in the last two months (a number that is higher than the number of actual devices, since most seem to have dynamic IP addresses). The ISPs behind those IP addresses have received notifications via email... Just curious... How well is that working out? For the IoT botnets, most of the emails are ignored or rejected, because most go to providers who either quietly bitbucket them or flat-out reject all abuse emails. Most emails sent to mainland China, for instance, are in that category (Hong Kong ISPs are somewhat better). For other botnets, such as those using compromised webservers running outdated phpMyAdmin installs at random hosts, harnessing spun-up services at reputable VPS providers (Amazon, Microsoft, Rackspace, etc.), or harnessing devices at large and small US and Canadian ISPs, we have had better luck. Usually, we don't hear a response back, but those emails are often forwarded to the end-user, who takes action (and may ask us for help, which is how we know they are being forwarded). The fixes can enough to reduce attack volumes to more manageable levels. Kudos go out to the large and small ISPs and NSPs who have started policing SSDP and other reflection traffic, which we also send out some notifications for. In some cases, it may be that our emails spurred them to notice how much damage those attacks were doing and how much it was costing them to carry the attack traffic. I've tried this myself a few times in the past, when I've found things that appear to be seriously compromised, and for my extensive trouble I've mostly received back utter silence and no action. I remember that after properly notifying security@ some large end-luser cable network in the SouthEast (which shall remain nameless) I got back something along the lines of "Thank you. We'll look into it." and was disgusted to find, two months later, that the boxes in question were still utterly pwned and in the exact same state they were two months prior, when I had first reported them. We do get our share of that, as well, unfortunately, along with our share of people who send angry responses calling the notifications spam (I disagree with them that sending a legitimate abuse notification to a publicly-posted, designated abuse account should be considered spam) or who flame us for acting like "internet police". But, we persist. Some people change their minds after receiving multiple notifications or after we explain that DoS traffic costs them money and hurts their customers, who will be experiencing degraded service and may silently switch providers over it. I guess that's just an example of what somebody else already noted here, i.e. that providers don't care to spend the time and/or effort and/or money necessary to actually -do- anything about compromised boxes, and anyway, they don't want to lose a paying customer. So, you know, let's just say for the sake of argument that right now, today, I know about a botnet consiting of a quarter million popped boxes, and that I have in-hand all of the relevant IPs, and that I have no trouble finding contact email addresses for all of the relevant ASNs. So then what? I use scripts to send out an abuse notification to some percentage of the compromised hosts -- the ones sending some significant amount of the traffic. The notification includes a description of what we saw and timestamped example attack traffic, as interpreted by tcpdump. If further traffic is seen later from the same host, another notification will be sent, after a cool-off period. The emails are plain text and we don't try to use them as advertisement. We also don't force a link to be clicked to see more details or to respond back. I don't like to receive such emails myself and have found that those types are more likely to be ignored. The question is: Why should I waste my time informing all, or even any of these ASNs about the popped boxes on their networks when (a) I am not their customer... as many of them have been only too happy to gleefully inform me in the past... and when (b) the vast majority simply won't do anything with the information? I'm not saying that everyone should send abuse notifications like we do, since it can be a big task. But, in response to someone wondering if their network is being used for attacks, or asking how they could help to police their own network, I am saying that making sure that inbound abuse notifications are arriving at the right place and being handled appropriately is important. And while we are on the subject, I just have to bring up one of my biggest pet peeves. Why is it that every time some public-spirited altrusitc well-meaning citizen such as myself reports any kind of a problem to any kind of a company on the Internet, the report itself gets immediately labeled and categorized as a "complaint".
Re: Death of the Internet, Film at 11
Ok, so this mailing list is a list of network operators. Swell. Every network operator who can do so, please raise your hand if you have *recently* scanned you own network and if you can -honestly- attest that you have taken all necessary steps to insure that none of the numerous specific types of CCVT thingies that Krebs and others identified weeks or months ago as being fundamentally insecure can emit a single packet out onto the public Internet. Most of the time, scanning of your customers isn't strictly necessary, though it certainly won't hurt. That's because attackers will scan your network /for /you, compromise the hosts, and use them to attack. When they inevitably attack one of my customers, I'll send you an abuse email. Some other networks do the same. So if you want to help, the real keys are to make sure that you disallow spoofing, that the RIR has up-to-date contact information for your organization, and that you handle abuse notifications effectively. Large IoT botnets have been used extensively this year, launching frequent 100+ Gbps attacks (they were also used in prior years, but it wasn't to the degree that we've seen since January 2016). I've recorded about 2.4 million IP addresses involved in the last two months (a number that is higher than the number of actual devices, since most seem to have dynamic IP addresses). The ISPs behind those IP addresses have received notifications via email, so if you haven't heard anything, you're probably in good shape, assuming the RIR has the right abuse address on file for you. The bulk of the compromised devices are non-NA. In a relatively small 40 Gbps IoT attack a couple of days ago, we saw about 20k devices, for instance, and most were from a mix of China, Brazil, Russia, Korea, and Venezuela. -John
Re: 20-30Gbps UDP 1720 traffic appearing to originate from CN in last 24 hours
Ca, Folks, it may be time to take the next step and admit that UDP is too broken to support https://tools.ietf.org/html/draft-byrne-opsec-udp-advisory-00 Your comments have been requested My comment would be that UDP is still widely used for game server traffic. This is unlikely to change in the near future because TCP (by default) is not well-suited for highly time-sensitive data, as even a small amount of packet loss causes significant delays. In light of this, it is a bad idea for network operators to apply overall rate-limits to UDP traffic right now. Rate-limiting specific UDP /ports/ that are frequently seen in reflection attacks -- such as 19, 123, and 1900 -- is a more reasonable practice, however, and it is becoming more common/. /UDP-based application protocols can be implemented correctly, such that they also have handshakes that limit their ability to be used for reflection attacks, and modern services (including modern game servers) do this. TCP and UDP can both be spoofed and used for direct attacks; we see this all the time. UDP is preferred due to many applications protocols' susceptibility to amplification attacks, but spoofed TCP attacks are often a bit thornier to deal with from the standpoint of a host attempting to externally mitigate, because tracking the three-way handshake requires keeping state. I spoke with Drew earlier and his attacks do not appear to be reflected, so this is orthogonal to his concern today. He is seeing directly-generated traffic, which could use any protocol. -John
Re: Cogent / Internap issue ??
On 5/27/2014 11:24 AM, Matthew Huff wrote: We are having troubles reaching services on the other side of cogent/internap peering. Anyone else seeing issues? We haven't seen Cogent-related issues at SEF today and that IP address is currently pingable through the Cogent looking glass. From your trace, Internap's equipment is reachable and responding (the two last hops are Internap-controlled), so the actual endpoint network (beyond Internap) seems to be either explicitly filtering your source or sending traffic back to it over a different and broken path. The Internap NOC is responsive when it comes to investigating routing problems, so they're a good place to turn for such concerns. If you're having problems reaching them or would like some external help in exploring the problem from inside and outside that PNAP, shoot me an email off-list. -John
Re: Filter NTP traffic by packet size?
On 2/20/2014 12:41 PM, Edward Roels wrote: Curious if anyone else thinks filtering out NTP packets above a certain packet size is a good or terrible idea. From my brief testing it seems 90 bytes for IPv4 and 110 bytes for IPv6 are typical for a client to successfully synchronize to an NTP server. If I query a server for it's list of peers (ntpq -np ip) I've seen packets as large as 522 bytes in a single packet in response to a 54 byte query. I'll admit I'm not 100% clear of the what is happening protocol-wise when I perform this query. I see there are multiple packets back forth between me and the server depending on the number of peers it has? Would I be breaking something important if I started to filter NTP packets 200 bytes into my network? If your equipment supports this, and you're seeing reflected NTP attacks, then it is an effective stopgap to block nearly all of the inbound attack traffic to affected hosts. Some still comes through from NTP servers running on nonstandard ports, but not much. Standard IPv4 NTP response packets are 76 bytes (plus any link-level headers), based on my testing. I have been internally filtering packets of other sizes against attack targets for some time now with no ill-effect. -John