Re: free collaborative tools for low BW and losy connections
> Exactly. And there's no disconnect: usenet doesn't scale because each object > is copied to all core nodes rather than referenced, or copied-as-needed, or > other. This design of distributed messaging platform will eventually break > as it grows. Usenet scales far more gracefully than the current web. Each node sends content to a few downstream nodes. This makes it easy to scale; there is no central mega-node that gets overwhelmed, connectivity is to a nearby upstream where there is a reasonabe amount of bandwidth. Last time I ran a server, the sender could filter based on newsgroup or message size, so avoid swamping links. Content was mostly text. It is possible to use offline transmission — certain groups dumped onto mag tape and mailed, get pulled in at the destination. BTDT. More demand = more client nodes which in turn distribute to other nodes, so each node does not need to talk to a large number of others. We did this about 30 years ago in South Africa; Rhodes university brought in most groups, I brought in alt.*. We each distributed to a select number of nodes, who distributed again. Lather, rinse, repeat. Usenet for the entire sub-continent (along with email) over 9600 bps dial-up circuits. paul
Re: free collaborative tools for low BW and losy connections
On 3/31/20 10:06 AM, Nick Hilliard wrote: Not pretty, but at least it could fit 4 xterms on-screen. In that sense, it was almost as functional as my ragingly fast desktop is these days. Link - Terminal forever <3 - http://www.commitstrip.com/en/2016/12/22/terminal-forever/ -- Grant. . . . unix || die smime.p7s Description: S/MIME Cryptographic Signature
Re: free collaborative tools for low BW and losy connections
Joe Greco wrote on 31/03/2020 15:55: There's a strange disconnect here. The concept behind Usenet is to have a distributed messaging platform. It isn't clear how this would work without ... well, distribution. The choice is between flood fill and perhaps something a little smarter, for which options were proposed and designed and never really caught on. Without the distribution mechanism (flooding), you don't have Usenet, you have something else entirely. Exactly. And there's no disconnect: usenet doesn't scale because each object is copied to all core nodes rather than referenced, or copied-as-needed, or other. This design of distributed messaging platform will eventually break as it grows. It's ok to acknowledge this explicitly: message buses are useful, but they have their limits. Kinda like how there's a problem with the technology of the Internet because if I wanna be a massive network or a tier 1 or whatever, I gotta have a massive investment in routers and 100G circuits and all that? Why can't we just build an Internet out of 10 megabit ethernet and T1's? Isn't this just another example of your "problem with the technology at the design level?" No, not even slightly. Is an NSP expected to carry all traffic for the entire DFZ? Because that's your proposed analogue here. Usage grows. I used to run Usenet on a 24MB Sun 3/60 with a pile of disks and Telebits. Now I'm blowing gigabits through massive machines. This isn't a poorly designed technology. It's scaled well past what anyone would have expected. yeah, I don't miss those days. I ran news on a decsystem 5100 with a couple of megs of RAM and a single disk. My desktop was a sun 3/60. Not pretty, but at least it could fit 4 xterms on-screen. In that sense, it was almost as functional as my ragingly fast desktop is these days. Nick
Re: free collaborative tools for low BW and losy connections
On Tue, Mar 31, 2020 at 01:46:09PM +0100, Nick Hilliard wrote: > Joe Greco wrote on 29/03/2020 23:14: > Flood often works fine until you attempt to scale it. Then it breaks, > just like Bj??rn admitted. Flooding is inherently problematic at scale. > >>> > >>>For... what, exactly? General Usenet? > >> > >>yes, this is what we're talking about. It couldn't scale to general > >>usenet levels. > > > >The scale issue wasn't flooding, it was bandwidth and storage. > > the bandwidth and storage problems happened because of flooding. Short > of cutting off content, there's no way to restrict bandwidth usage, but > cutting off content restricts the functionality of the ecosystem. You > can work around this using remote readers and manually distributing , > but there's still a fundamental scaling issue going on here, namely that > the model of flooding all posts in all groups to all nodes has terrible > scaling design characteristics. It's terrible because it requires all > core nodes to linearly scale their individual resourcing requirements > according to the overall load of the entire system. You can manually > configure load splitting to work around some of these limitations, but > it's not possible to ignore the design problems here. There's a strange disconnect here. The concept behind Usenet is to have a distributed messaging platform. It isn't clear how this would work without ... well, distribution. The choice is between flood fill and perhaps something a little smarter, for which options were proposed and designed and never really caught on. Without the distribution mechanism (flooding), you don't have Usenet, you have something else entirely. > [...] > >The Usenet "backbone" with binaries isn't going to be viable without a > >real large capex investment and significant ongoing opex. This isn't a > >failure in the technology. > > We may need to agree to disagree on this then. Reasonable engineering > entails being able to build workable solutions within a feasible budget. > If you can't do this, then there's a problem with the technology at > the design level. Kinda like how there's a problem with the technology of the Internet because if I wanna be a massive network or a tier 1 or whatever, I gotta have a massive investment in routers and 100G circuits and all that? Why can't we just build an Internet out of 10 megabit ethernet and T1's? Isn't this just another example of your "problem with the technology at the design level?" See, here's the thing. Twenty six years ago, one of the local NSP's here spent some modest thousands of dollars on a few routers, a switch, and a circuit down to Chicago and set up shop on a folding table (really). This was not an unreasonable outlay of cash to get bootstrapped back in those days. However, within just a few years, the amount of cash that you'd need to invest to get started as an NSP had exploded dramatically. Usage grows. I used to run Usenet on a 24MB Sun 3/60 with a pile of disks and Telebits. Now I'm blowing gigabits through massive machines. This isn't a poorly designed technology. It's scaled well past what anyone would have expected. > >Usenet is a great technology for doing collaboration on low bandwidth and > >lossy connections. > > For small, constrained quantities of traffic, it works fine. It seems like that was the point of this thread... ... JG -- Joe Greco - sol.net Network Services - Milwaukee, WI - http://www.sol.net "The strain of anti-intellectualism has been a constant thread winding its way through our political and cultural life, nurtured by the false notion that democracy means that 'my ignorance is just as good as your knowledge.'"-Asimov
Re: free collaborative tools for low BW and losy connections
Joe Greco wrote on 29/03/2020 23:14: Flood often works fine until you attempt to scale it. Then it breaks, just like Bj??rn admitted. Flooding is inherently problematic at scale. For... what, exactly? General Usenet? yes, this is what we're talking about. It couldn't scale to general usenet levels. The scale issue wasn't flooding, it was bandwidth and storage. the bandwidth and storage problems happened because of flooding. Short of cutting off content, there's no way to restrict bandwidth usage, but cutting off content restricts the functionality of the ecosystem. You can work around this using remote readers and manually distributing , but there's still a fundamental scaling issue going on here, namely that the model of flooding all posts in all groups to all nodes has terrible scaling design characteristics. It's terrible because it requires all core nodes to linearly scale their individual resourcing requirements according to the overall load of the entire system. You can manually configure load splitting to work around some of these limitations, but it's not possible to ignore the design problems here. [...] The Usenet "backbone" with binaries isn't going to be viable without a real large capex investment and significant ongoing opex. This isn't a failure in the technology. We may need to agree to disagree on this then. Reasonable engineering entails being able to build workable solutions within a feasible budget. If you can't do this, then there's a problem with the technology at the design level. Usenet is a great technology for doing collaboration on low bandwidth and lossy connections. For small, constrained quantities of traffic, it works fine. Nick
Re: free collaborative tools for low BW and losy connections
On 3/30/20 11:18 AM, Keith Medcalf wrote: On Monday, 30 March, 2020 11:19, Michael Thomas wrote: On 3/30/20 5:52 AM, Rich Kulawiec wrote: On Mon, Mar 30, 2020 at 06:30:16AM -0500, Joe Greco wrote: Actual text traffic has been slowly dying off for years as webforums have matured and become a better choice of technology for nontechnical end users on high speed Internet connections. My view is that the move to web forums is a huge downgrade. Mailing lists are vastly superior. [] The thing that mailing lists lack is a central directory of their existence. The discovery problem is a pretty big one. Where is this to be found for webforums? I have never seen one. Or do you think Google is such a master index? Can you please pose your Google query that you think results in a comprehensive index of *all* webforums? Or is your comment nothing more that you noticing that NEITHER e-mail lists NOR webforums have a master index, which is a rather useless observation that would indicate that webforums have zero advantage over mailing lists in this regard, so what is the point of the whataboutism? I didn't expect to get an apology, and wasn't disappointed. MIke
Re: free collaborative tools for low BW and losy connections
On Mon, Mar 30, 2020 at 8:56 AM Rich Kulawiec wrote: > On Mon, Mar 30, 2020 at 06:30:16AM -0500, Joe Greco wrote: > > Actual text traffic has been slowly dying off for years as webforums > > have matured and become a better choice of technology for nontechnical > > end users on high speed Internet connections. > > My view is that the move to web forums is a huge downgrade. Mailing lists > are vastly superior. > Are web forums even still much of a thing in recent years? My own experience in several non-networking realms, where I was active in a number of web-based forums, is that over the past 4 or 5 years facebook groups, both public and private, have siphoned off the bulk of the former discussion traffic of once-thriving web forums, with few exceptions. Talk about a huge downgrade. While facebook's groups allow for virtually unlimited image uploads, the are extremely lacking in features such as threaded discussion and searching. I grudgingly live in a number of facebook groups, including one I admin, but only because all the people I know from usenet, and then later in forums, have migrated to facebook groups. One popular city-based discussion group withered away from hundreds of posts and comments daily to sometimes several days with NO comments at all. Network effect is in full effect.
Re: free collaborative tools for low BW and losy connections
On 3/30/20 11:28 AM, Joe Greco wrote: On Mon, Mar 30, 2020 at 12:18:37PM -0600, Keith Medcalf wrote: The thing that mailing lists lack is a central directory of their existence. The discovery problem is a pretty big one. Where is this to be found for webforums? I have never seen one. Or do you think Google is such a master index? Can you please pose your Google query that you think results in a comprehensive index of *all* webforums? Or is your comment nothing more that you noticing that NEITHER e-mail lists NOR webforums have a master index, which is a rather useless observation that would indicate that webforums have zero advantage over mailing lists in this regard, so what is the point of the whataboutism? In the context of the discussion, I expect the point is that Usenet had such functionality. This is a significant feature advantage. As are a bunch of other things pointed out earlier today. Even a steaming piece of crap software like Reddit has search facilities. They may be buggy, but they're there. Nothing I know of that's even remotely similar for email lists. Mike
Re: free collaborative tools for low BW and losy connections
On Mon, Mar 30, 2020 at 12:18:37PM -0600, Keith Medcalf wrote: > >The thing that mailing lists lack is a central directory of their > >existence. The discovery problem is a pretty big one. > > Where is this to be found for webforums? I have never seen one. Or do > you think Google is such a master index? Can you please pose your > Google query that you think results in a comprehensive index of *all* > webforums? > > Or is your comment nothing more that you noticing that NEITHER e-mail > lists NOR webforums have a master index, which is a rather useless > observation that would indicate that webforums have zero advantage > over mailing lists in this regard, so what is the point of the > whataboutism? In the context of the discussion, I expect the point is that Usenet had such functionality. This is a significant feature advantage. As are a bunch of other things pointed out earlier today. ... JG -- Joe Greco - sol.net Network Services - Milwaukee, WI - http://www.sol.net "The strain of anti-intellectualism has been a constant thread winding its way through our political and cultural life, nurtured by the false notion that democracy means that 'my ignorance is just as good as your knowledge.'"-Asimov
Re: free collaborative tools for low BW and losy connections
On 3/30/20 11:18 AM, Keith Medcalf wrote: On Monday, 30 March, 2020 11:19, Michael Thomas wrote: On 3/30/20 5:52 AM, Rich Kulawiec wrote: On Mon, Mar 30, 2020 at 06:30:16AM -0500, Joe Greco wrote: Actual text traffic has been slowly dying off for years as webforums have matured and become a better choice of technology for nontechnical end users on high speed Internet connections. My view is that the move to web forums is a huge downgrade. Mailing lists are vastly superior. [] The thing that mailing lists lack is a central directory of their existence. The discovery problem is a pretty big one. Where is this to be found for webforums? I have never seen one. Or do you think Google is such a master index? Can you please pose your Google query that you think results in a comprehensive index of *all* webforums? Or is your comment nothing more that you noticing that NEITHER e-mail lists NOR webforums have a master index, which is a rather useless observation that would indicate that webforums have zero advantage over mailing lists in this regard, so what is the point of the whataboutism? Usenet had them. Mike
RE: free collaborative tools for low BW and losy connections
On Monday, 30 March, 2020 11:19, Michael Thomas wrote: >On 3/30/20 5:52 AM, Rich Kulawiec wrote: >> On Mon, Mar 30, 2020 at 06:30:16AM -0500, Joe Greco wrote: >>> Actual text traffic has been slowly dying off for years as webforums >>> have matured and become a better choice of technology for nontechnical >>> end users on high speed Internet connections. >> My view is that the move to web forums is a huge downgrade. Mailing >> lists are vastly superior. >[] >The thing that mailing lists lack is a central directory of their >existence. The discovery problem is a pretty big one. Where is this to be found for webforums? I have never seen one. Or do you think Google is such a master index? Can you please pose your Google query that you think results in a comprehensive index of *all* webforums? Or is your comment nothing more that you noticing that NEITHER e-mail lists NOR webforums have a master index, which is a rather useless observation that would indicate that webforums have zero advantage over mailing lists in this regard, so what is the point of the whataboutism? -- The fact that there's a Highway to Hell but only a Stairway to Heaven says a lot about anticipated traffic volume.
Re: free collaborative tools for low BW and losy connections
On 3/30/20 5:52 AM, Rich Kulawiec wrote: On Mon, Mar 30, 2020 at 06:30:16AM -0500, Joe Greco wrote: Actual text traffic has been slowly dying off for years as webforums have matured and become a better choice of technology for nontechnical end users on high speed Internet connections. My view is that the move to web forums is a huge downgrade. Mailing lists are vastly superior. [] The thing that mailing lists lack is a central directory of their existence. The discovery problem is a pretty big one. Mike
Re: free collaborative tools for low BW and losy connections
On Mon, Mar 30, 2020 at 06:30:16AM -0500, Joe Greco wrote: > Actual text traffic has been slowly dying off for years as webforums > have matured and become a better choice of technology for nontechnical > end users on high speed Internet connections. My view is that the move to web forums is a huge downgrade. Mailing lists are vastly superior. Here's a (large) snip from http://www.firemountain.net/why-mailing-lists.html (which I wrote) with a comment from me shoved in the middle. [...] 1. Mailing lists require no special software: anyone with a sensible mail client can participate. Thus they allow you to use *your* software with the user interface of *your* choosing rather than being compelled to learn 687 different web forums with 687 different user interfaces, all of which range from "merely bad" to "hideously bad". 2. Mailing lists are simple: learn a few basic rules of netiquette and a couple of Internet-wide conventions, and one's good to go. Web forums are complicated because all of them are different. In other words, participating in 20 different mailing lists is just about as easy as participating in one; but participating in 20 different web forums is onerous. 3. They impose minimal security risk. 4. They impose minimal privacy risk. Points 3 and 4 stand in stark contrast to the security and privacy risks imposed on users of web forums and "social" media, especially the latter. 5. Mailing lists are bandwidth-friendly -- an increasing concern for people on mobile devices and thus on expensive data plans. Web forums are bandwidth-hungry. 6. Mailing lists interoperate. I can easily forward a message from one list to another one. Or to a person. I can send a message to multiple lists. I can forward a message from a person to this list. And so on. Try doing this with web forum software A on host B with destinations web forum software X and Y on hosts X1 and Y1. Good luck with that. 7. They're asynchronous: you don't have to interact in real time. You can download messages when connected to the Internet, then read them and compose responses when offline. 8. As a result of 7, They work reasonably well even in the presence of multiple outages and severe congestion. Messages may be delayed, but once everything's up again, they'll go through. Web-based forums simply don't work. 9. They're push, not pull, so new content just shows up. Web forums require that you go fishing for it. 10. They scale beautifully. 11. Mailing lists -- when properly run -- are highly resistant to abuse. Web forums, because of their complexity, are highly vulnerable to software security issues as well as spam/phishing and other attacks. [ I'm going to interject a comment here that's not on the web page I'm quoting myself from. There are, of course, counter examples to this. There is a very busy very well-known mailing list that is an absolute cesspool of trivially-blockable spam. Hence the phrase "when properly run", becase when that's done spam incidents should be at the 1 per year level or less. It's not that hard.] 12. They handle threading well. And provided users take a few seconds to edit properly, they handle quoting well. 13. They're portable: lists can be rehosted (different domain, different host) rather easily. 14. They can be freely interconverted -- that is, you can move a list hosted by A using software B on operating system C to host X using software Y on operating system Z. 15. They can be written to media and read from it. This is a very non-trivial task with web forums. 16. The computing resources require to support them are minimal -- CPU, memory, disk, bandwidth, etc. 17. Mailing lists can be uni- or bidirectionally gatewayed to Usenet. (The main Python language mailing list is an example of this.) This can be highly useful. 18. They're easily archivable in a format that is simple and likely to be readable long into the future. Mail archives from 10, 20, even 30 or more years ago are still completely usable. And they take up very little space. (Numerous tools exist for handling Unix "mbox" format: for example, "grepmail" is a highly useful basic search tool. Most search engines include parsers for email, and the task of ingesting mail archives into search engines is very well understood.) 19. You can archive them locally... 20. ...which means you can search them locally with the software of *your* choice. Including when you're offline. And provided you make backups, you'll always have an archive -- even if the original goes away. Web forums don't facilitate this. (Those of who've been around for a while have seen a lot of web-based discussions vanish forever because a host crashed or a domain expired or a company went under or a company was acquired or someone made a mistake or there was a security breach or a government confiscated it.) [...] ---rsk
Re: free collaborative tools for low BW and losy connections
> > Actual text traffic has been slowly dying off for years as webforums > have matured and become a better choice of technology for nontechnical > end users on high speed Internet connections. > The solaris groups for info worked great Psinet uk liked using the sun kit for great nntp even uucp nntp as well Col
Re: free collaborative tools for low BW and losy connections
On Sun, Mar 29, 2020 at 04:18:51PM -0700, Michael Thomas wrote: > > On 3/29/20 1:46 PM, Joe Greco wrote: > >On Sun, Mar 29, 2020 at 07:46:28PM +0100, Nick Hilliard wrote: > >>Joe Greco wrote on 29/03/2020 15:56: > >> > >>The concept of flooding isn't problematic by itself. > >>Flood often works fine until you attempt to scale it. Then it breaks, > >>just like Bj??rn admitted. Flooding is inherently problematic at scale. > >For... what, exactly? General Usenet? Perhaps, but mainly because you > >do not have a mutual agreement on traffic levels and a bunch of other > >factors. Flooding works just fine within private hierarchies, and since > >I thought this was a discussion of "free collaborative tools" rather than > >"random newbie trying to masochistically keep up with a full backbone > >Usenet feed", it definitely should work fine for a private hierarchy and > >collaborative use. > > AFAIK, Usenet didn't die because it wasn't scalable. It died because > people figured out how to make it a business model. Not at all. I can see why you say that, but it isn't the reality, any more than commercial uses killed the Internet when it was opened up to people who made it a business model. The introduction of the DMCA ratcheted up the potential for enforcement and penalties against end users doing Napster, bittorrent, illicit web and FTP, or whatever your other favorite form of digital piracy might happen to have been back in the '90's. The CDA 230 protection for providers allowed Usenet to be served without significant concern. So pirates had a safe model where they could distribute pirated traffic, posting it somewhere "safe" and then it would be available everywhere, and it was HARD to get it taken down. This, along with legitimate binaries traffic increases, caused an explosion in traffic, which made Usenet increasingly impractical for ISP's to self-host. The problem is that it scaled far too well as a binary traffic distribution system. As this happened, most ISP's outsourced to Usenet service providers, and end users often picked up "premium" Usenet services from such providers directly as well. I do not see the people who made it a business model as responsible for the state of affairs. Had commercial USP's not stepped up, Usenet probably would have died off in the late 90's-early 2000's as ISP's dropped support. They (and I have to include myself as I run a Usenet company) are arguably the ones who kept it going. Demand was there. The users who are dumping binaries on Usenet are helping to kill it. Actual text traffic has been slowly dying off for years as webforums have matured and become a better choice of technology for nontechnical end users on high speed Internet connections. ... JG -- Joe Greco - sol.net Network Services - Milwaukee, WI - http://www.sol.net "The strain of anti-intellectualism has been a constant thread winding its way through our political and cultural life, nurtured by the false notion that democracy means that 'my ignorance is just as good as your knowledge.'"-Asimov
Re: free collaborative tools for low BW and losy connections
On 3/29/20 1:46 PM, Joe Greco wrote: On Sun, Mar 29, 2020 at 07:46:28PM +0100, Nick Hilliard wrote: Joe Greco wrote on 29/03/2020 15:56: The concept of flooding isn't problematic by itself. Flood often works fine until you attempt to scale it. Then it breaks, just like Bj??rn admitted. Flooding is inherently problematic at scale. For... what, exactly? General Usenet? Perhaps, but mainly because you do not have a mutual agreement on traffic levels and a bunch of other factors. Flooding works just fine within private hierarchies, and since I thought this was a discussion of "free collaborative tools" rather than "random newbie trying to masochistically keep up with a full backbone Usenet feed", it definitely should work fine for a private hierarchy and collaborative use. AFAIK, Usenet didn't die because it wasn't scalable. It died because people figured out how to make it a business model. Mike
Re: free collaborative tools for low BW and losy connections
On Sun, Mar 29, 2020 at 07:46:28PM +0100, Nick Hilliard wrote: > Joe Greco wrote on 29/03/2020 15:56: > >On Sun, Mar 29, 2020 at 03:01:04PM +0100, Nick Hilliard wrote: > >>because it uses flooding and can't guarantee reliable message > >>distribution, particularly at higher traffic levels. > > > >That's so hideously wrong. It's like claiming web forums don't > >work because IP packet delivery isn't reliable. > > Really, it's nothing like that. Sure it is. At a certain point you can get web forums to stop working by DDoS. You can't guarantee reliable interaction with a web site if that happens. > >Usenet message delivery at higher levels works just fine, except that > >on the public backbone, it is generally implemented as "best effort" > >rather than a concerted effort to deliver reliably. > > If you can explain the bit of the protocol that guarantees that all > nodes have received all postings, then let's discuss it. There isn't, just like there isn't a bit of the protocol that guarantees that an IP packet is received by its intended recipient. No magic. It's perfectly possible to make sure that you are not backlogging to a peer and to contact them to remediate if there is a problem. When done at scale, this does actually work. And unlike IP packet delivery, news will happily backlog and recover from a server being down or whatever. > >The concept of flooding isn't problematic by itself. > > Flood often works fine until you attempt to scale it. Then it breaks, > just like Bj??rn admitted. Flooding is inherently problematic at scale. For... what, exactly? General Usenet? Perhaps, but mainly because you do not have a mutual agreement on traffic levels and a bunch of other factors. Flooding works just fine within private hierarchies, and since I thought this was a discussion of "free collaborative tools" rather than "random newbie trying to masochistically keep up with a full backbone Usenet feed", it definitely should work fine for a private hierarchy and collaborative use. > > If you wanted to > >implement a collaborative system, you could easily run a private > >hierarchy and run a separate feed for it, which you could then monitor > >for backlogs or issues. You do not need to dump your local traffic on > >the public Usenet. This can happily coexist alongside public traffic > >on your server. It is easy to make it 100% reliable if that is a goal. > > For sure, you can operate mostly reliable self-contained systems with > limited distribution. We're all in agreement about this. Okay, good. > >>The fact that it ended up having to implement TAKETHIS is only one > >>indication of what a truly awful protocol it is. > > > >No, the fact that it ended up having to implement TAKETHIS is a nod to > >the problem of RTT. > > TAKETHIS was necessary to keep things running because of the dual > problem of RTT and lack of pipelining. Taken together, these two > problems made it impossible to optimise incoming feeds, because of ... > well, flooding, which meant that even if you attempted an IHAVE, by the > time you delivered the article, some other feed might already have > delivered it. TAKETHIS managed to sweep these problems under the > carpet, but it's a horrible, awful protocol hack. It's basically cheap pipelining. If you want to call pipelining in general a horrible, awful protocol hack, then that's probably got some validity. > >It did and has. The large scale binaries sites are still doing a > >great job of propagating binaries with very close to 100% reliability. > > which is mostly because there are so few large binary sites these days, > i.e. limited distribution model. No, there are so few large binary sites these days because of consolidation and buyouts. > >I was there. > > So was I, and probably so were lots of other people on nanog-l. We all > played our part trying to keep the thing hanging together. I'd say most of the folks here were out of this fifteen to twenty years ago, well before the explosion of binaries in the early 2000's. > >I'm the maintainer of Diablo. It's fair to say I had a > >large influence on this issue as it was Diablo's distributed backend > >capability that really instigated retention competition, and a number > >of optimizations that I made helped make it practical. > > Diablo was great - I used it for years after INN-related head-bleeding. > Afterwards, Typhoon improved things even more. > > >The problem for smaller sites is simply the immense traffic volume. > >If you want to carry binaries, you need double digits Gbps. If you > >filter them out, the load is actually quite trivial. > > Right, so you've put your finger on the other major problem relating to > flooding which isn't the distribution synchronisation / optimisation > problem: all sites get all posts for all groups which they're configured > for. This is a profound waste of resources + it doesn't scale in any > meaningful way. So if you don't like that
Re: free collaborative tools for low BW and losy connections
On Sun, Mar 29, 2020 at 03:01:04PM +0100, Nick Hilliard wrote: > Bj??rn Mork wrote on 29/03/2020 13:44: > >How is nntp non-scalable? > > because it uses flooding and can't guarantee reliable message > distribution, particularly at higher traffic levels. That's so hideously wrong. It's like claiming web forums don't work because IP packet delivery isn't reliable. Usenet message delivery at higher levels works just fine, except that on the public backbone, it is generally implemented as "best effort" rather than a concerted effort to deliver reliably. The concept of flooding isn't problematic by itself. If you wanted to implement a collaborative system, you could easily run a private hierarchy and run a separate feed for it, which you could then monitor for backlogs or issues. You do not need to dump your local traffic on the public Usenet. This can happily coexist alongside public traffic on your server. It is easy to make it 100% reliable if that is a goal. > The fact that it ended up having to implement TAKETHIS is only one > indication of what a truly awful protocol it is. No, the fact that it ended up having to implement TAKETHIS is a nod to the problem of RTT. > Once again in simpler terms: > > > How is nntp non-scalable? > [...] > > Binaries broke USENET. That has little to do with nntp. > > If it had been scalable, it could have scaled to handling the binary groups. It did and has. The large scale binaries sites are still doing a great job of propagating binaries with very close to 100% reliability. I was there. I'm the maintainer of Diablo. It's fair to say I had a large influence on this issue as it was Diablo's distributed backend capability that really instigated retention competition, and a number of optimizations that I made helped make it practical. The problem for smaller sites is simply the immense traffic volume. If you want to carry binaries, you need double digits Gbps. If you filter them out, the load is actually quite trivial. ... JG -- Joe Greco - sol.net Network Services - Milwaukee, WI - http://www.sol.net "The strain of anti-intellectualism has been a constant thread winding its way through our political and cultural life, nurtured by the false notion that democracy means that 'my ignorance is just as good as your knowledge.'"-Asimov
Re: free collaborative tools for low BW and losy connections
On Sun, Mar 29, 2020 at 10:31:50PM +0100, Nick Hilliard wrote: > Joe Greco wrote on 29/03/2020 21:46: > >On Sun, Mar 29, 2020 at 07:46:28PM +0100, Nick Hilliard wrote: > >>>That's so hideously wrong. It's like claiming web forums don't > >>>work because IP packet delivery isn't reliable. > >> > >>Really, it's nothing like that. > > > >Sure it is. At a certain point you can get web forums to stop working > >by DDoS. You can't guarantee reliable interaction with a web site if > >that happens. > > this is failure caused by external agency, not failure caused by > inherent protocol limitations. Yet we're discussing "low BW and losy(sic) connections". Which would be failure of IP to be magically available with zero packet loss and at high speeds. There are lots of people for whom low speed DSL, dialup, WISP, 4G, GPRS, satellite, or actually nothing at all are available as the Internet options. > >>>Usenet message delivery at higher levels works just fine, except that > >>>on the public backbone, it is generally implemented as "best effort" > >>>rather than a concerted effort to deliver reliably. > >> > >>If you can explain the bit of the protocol that guarantees that all > >>nodes have received all postings, then let's discuss it. > > > >There isn't, just like there isn't a bit of the protocol that guarantees > >that an IP packet is received by its intended recipient. No magic. > > tcp vs udp. IP vs ... what exactly? > >>Flood often works fine until you attempt to scale it. Then it breaks, > >>just like Bj??rn admitted. Flooding is inherently problematic at scale. > > > >For... what, exactly? General Usenet? > > yes, this is what we're talking about. It couldn't scale to general > usenet levels. The scale issue wasn't flooding, it was bandwidth and storage. It's actually not problematic to do history lookups (the key mechanism in what you're calling "flooding") because even at a hundred thousand per second, that's well within the speed of CPU and RAM. Oh, well, yes, if you're trying to do it on HDD, that won't work anymore, and quite possibly SSD will reach limits. But that's a design issue, not a scale problem. Most of Usenet's so-called "scale" problems had to do with disk I/O and network speeds, not flood fill. > >Perhaps, but mainly because you > >do not have a mutual agreement on traffic levels and a bunch of other > >factors. Flooding works just fine within private hierarchies and since > >I thought this was a discussion of "free collaborative tools" rather than > >"random newbie trying to masochistically keep up with a full backbone > >Usenet feed", it definitely should work fine for a private hierarchy and > >collaborative use. > > Then we're in violent agreement on this point. Great! Okay, fine, but it's kinda the same thing as "last week some noob got a 1990's era book on setting up a webhost, bought a T1, and was flummoxed at why his service sucked." The Usenet "backbone" with binaries isn't going to be viable without a real large capex investment and significant ongoing opex. This isn't a failure in the technology. > >>delivered it. TAKETHIS managed to sweep these problems under the > >>carpet, but it's a horrible, awful protocol hack. > > > >It's basically cheap pipelining. > > no, TAKETHIS is unrestrained flooding, not cheap pipelining. It is definitely not unrestrained. Sorry, been there inside the code. There's a limited window out of necessity, because you get interesting behaviours if a peer is held off too long. > >If you want to call pipelining in > >general a horrible, awful protocol hack, then that's probably got > >some validity. > > you could characterise pipelining as a necessary reaction to the fact > that the speed of light is so damned slow. Sure. > >>which is mostly because there are so few large binary sites these days, > >>i.e. limited distribution model. > > > >No, there are so few large binary sites these days because of consolidation > >and buyouts. > > 20 years ago, lots of places hosted binaries. They stopped because it > was pointless and wasteful, not because of consolidation. I thought they stopped it because some of us offered them a better model that reduced their expenses and eliminated the need to have someone who was an expert in an esoteric '80's era service, while also investing in all the capex/opex. Lots of companies sold wholesale Usenet, usually just by offering access to a remote service. As the amount of Usenet content exploded, the increasing cost of storage for a feature a declining number of users was using didn't make sense. One of my companies specialized in shipping dreader boxes to ISP's and letting them backend off remote spools, usually for a fairly modest cost (high three, low four figures?). This let them have control over the service that was unlike anything other service providers were doing. Custom groups, custom integration with their auth/billing, etc., required about a megabit of
Re: free collaborative tools for low BW and losy connections
Joe Greco wrote on 29/03/2020 21:46: On Sun, Mar 29, 2020 at 07:46:28PM +0100, Nick Hilliard wrote: That's so hideously wrong. It's like claiming web forums don't work because IP packet delivery isn't reliable. Really, it's nothing like that. Sure it is. At a certain point you can get web forums to stop working by DDoS. You can't guarantee reliable interaction with a web site if that happens. this is failure caused by external agency, not failure caused by inherent protocol limitations. Usenet message delivery at higher levels works just fine, except that on the public backbone, it is generally implemented as "best effort" rather than a concerted effort to deliver reliably. If you can explain the bit of the protocol that guarantees that all nodes have received all postings, then let's discuss it. There isn't, just like there isn't a bit of the protocol that guarantees that an IP packet is received by its intended recipient. No magic. tcp vs udp. Flood often works fine until you attempt to scale it. Then it breaks, just like Bj??rn admitted. Flooding is inherently problematic at scale. For... what, exactly? General Usenet? yes, this is what we're talking about. It couldn't scale to general usenet levels. Perhaps, but mainly because you do not have a mutual agreement on traffic levels and a bunch of other factors. Flooding works just fine within private hierarchies and since I thought this was a discussion of "free collaborative tools" rather than "random newbie trying to masochistically keep up with a full backbone Usenet feed", it definitely should work fine for a private hierarchy and collaborative use. Then we're in violent agreement on this point. Great! delivered it. TAKETHIS managed to sweep these problems under the carpet, but it's a horrible, awful protocol hack. It's basically cheap pipelining. no, TAKETHIS is unrestrained flooding, not cheap pipelining. If you want to call pipelining in general a horrible, awful protocol hack, then that's probably got some validity. you could characterise pipelining as a necessary reaction to the fact that the speed of light is so damned slow. which is mostly because there are so few large binary sites these days, i.e. limited distribution model. No, there are so few large binary sites these days because of consolidation and buyouts. 20 years ago, lots of places hosted binaries. They stopped because it was pointless and wasteful, not because of consolidation. Right, so you've put your finger on the other major problem relating to flooding which isn't the distribution synchronisation / optimisation problem: all sites get all posts for all groups which they're configured for. This is a profound waste of resources + it doesn't scale in any meaningful way. So if you don't like that everyone gets everything they are configured to get, you are suggesting that they... what, exactly? Shouldn't get everything they want? The default distribution model of the 1990s was *. These days, only a tiny handful of sites handle everything, because the overheads of flooding are so awful. To make it clear, this awfulness is resource related, and the knock-on effect is that the resource cost is untenable. Usenet, like other systems, can be reduced to an engineering / economics management problem. If the cost of making it operate correctly doesn't work, then it's non-viable. None of this changes that it's a robust, mature protocol that was originally designed for handling non-binaries and is actually pretty good in that role. Having the content delivered to each site means that there is no dependence on long-distance interactive IP connections and that each participating site can keep the content for however long they deem useful. Usenet keeps hummin' along under conditions that would break more modern things like web forums. It's a complete crock of a protocol with robust and mature implementations. Diablo is one and for that, we have people like Matt and you to thank. Nick
Re: free collaborative tools for low BW and losy connections
Joe Greco wrote on 29/03/2020 15:56: On Sun, Mar 29, 2020 at 03:01:04PM +0100, Nick Hilliard wrote: because it uses flooding and can't guarantee reliable message distribution, particularly at higher traffic levels. That's so hideously wrong. It's like claiming web forums don't work because IP packet delivery isn't reliable. Really, it's nothing like that. Usenet message delivery at higher levels works just fine, except that on the public backbone, it is generally implemented as "best effort" rather than a concerted effort to deliver reliably. If you can explain the bit of the protocol that guarantees that all nodes have received all postings, then let's discuss it. The concept of flooding isn't problematic by itself. Flood often works fine until you attempt to scale it. Then it breaks, just like Bjørn admitted. Flooding is inherently problematic at scale. If you wanted to implement a collaborative system, you could easily run a private hierarchy and run a separate feed for it, which you could then monitor for backlogs or issues. You do not need to dump your local traffic on the public Usenet. This can happily coexist alongside public traffic on your server. It is easy to make it 100% reliable if that is a goal. For sure, you can operate mostly reliable self-contained systems with limited distribution. We're all in agreement about this. The fact that it ended up having to implement TAKETHIS is only one indication of what a truly awful protocol it is. No, the fact that it ended up having to implement TAKETHIS is a nod to the problem of RTT. TAKETHIS was necessary to keep things running because of the dual problem of RTT and lack of pipelining. Taken together, these two problems made it impossible to optimise incoming feeds, because of ... well, flooding, which meant that even if you attempted an IHAVE, by the time you delivered the article, some other feed might already have delivered it. TAKETHIS managed to sweep these problems under the carpet, but it's a horrible, awful protocol hack. It did and has. The large scale binaries sites are still doing a great job of propagating binaries with very close to 100% reliability. which is mostly because there are so few large binary sites these days, i.e. limited distribution model. I was there. So was I, and probably so were lots of other people on nanog-l. We all played our part trying to keep the thing hanging together. I'm the maintainer of Diablo. It's fair to say I had a large influence on this issue as it was Diablo's distributed backend capability that really instigated retention competition, and a number of optimizations that I made helped make it practical. Diablo was great - I used it for years after INN-related head-bleeding. Afterwards, Typhoon improved things even more. The problem for smaller sites is simply the immense traffic volume. If you want to carry binaries, you need double digits Gbps. If you filter them out, the load is actually quite trivial. Right, so you've put your finger on the other major problem relating to flooding which isn't the distribution synchronisation / optimisation problem: all sites get all posts for all groups which they're configured for. This is a profound waste of resources + it doesn't scale in any meaningful way. Nick
Re: free collaborative tools for low BW and losy connections
Bjørn Mork wrote on 29/03/2020 13:44: How is nntp non-scalable? because it uses flooding and can't guarantee reliable message distribution, particularly at higher traffic levels. The fact that it ended up having to implement TAKETHIS is only one indication of what a truly awful protocol it is. Once again in simpler terms: > How is nntp non-scalable? [...] > Binaries broke USENET. That has little to do with nntp. If it had been scalable, it could have scaled to handling the binary groups. Nick
Re: free collaborative tools for low BW and losy connections
Nick Hilliard writes: > nntp is a non-scalable protocol which broke under its own > weight. How is nntp non-scalable? It allows an infinite number of servers connected in a tiered network, where you only have to connect to a few other peers and carry whatever part of the traffic you want. Binaries broke USENET. That has little to do with nntp. nntp is still working just fine and still carrying a few discussion groups here and there. And you have a really nice mailling list gateway at news.gmane.io (which recently replaced gmane.org - see https://lars.ingebrigtsen.no/2020/01/15/news-gmane-org-is-now-news-gmane-io/ for full story) Bjørn
Re: free collaborative tools for low BW and losy connections
On Wed, Mar 25, 2020 at 05:27:41PM +, Nick Hilliard wrote: > nntp is a non-scalable protocol which broke under its own weight. Threaded > news-readers are a great way of catching up with large mailing lists if > you're prepared to put in the effort to create a bidirectional gateway. But > that's really a statement that mail readers are usually terrible at handling > large threads rather than a statement about nntp as a useful media delivery > protocol. Some mail readers are terrible at that: mutt isn't. And one of the nice things about trn (and I believe slrn, although that's an educated guess, I haven't checked) is that it can save Usenet news articles in Unix mbox format, which means that you can read them with mutt as well. I have trn set up to run via a cron job that executes a script that grabs the appropriate set of newsgroups, spam-filters them, saves what's left to a per-newsgroup mbox file that I can read just like I read this list. Similarly, rss2email saves RSS feeds in Unix mbox format. And one of the *very* nice things about coercing everything into mbox format is that myriad tools existing for sorting, searching, indexing, etc. ---rsk
Re: free collaborative tools for low BW and losy connections
On 3/25/20 11:27 AM, Nick Hilliard wrote: nntp is a non-scalable protocol which broke under its own weight. That statement surprises me. But I'm WAY late to the NNTP / Usenet game. Threaded news-readers are a great way of catching up with large mailing lists if you're prepared to put in the effort to create a bidirectional gateway. But that's really a statement that mail readers are usually terrible at handling large threads rather than a statement about nntp as a useful media delivery protocol. Especially when most of the news readers that I use or hear others talk about using are primarily email clients that also happen to be news clients. As such, it's the same threading code. -- Grant. . . . unix || die smime.p7s Description: S/MIME Cryptographic Signature
Re: free collaborative tools for low BW and losy connections
On 3/25/20 3:47 PM, Randy Bush wrote: some of us still do uucp, over tcp and over pots. My preference is to do UUCP over SSH (STDIO) over TCP/IP. IMHO the SSH adds security (encryption and more friendly authentication (keys / certs / Kerberos)) and reduces the number of ports that need to be exposed to the world / allowed through the network. archaic, but still the right tool for some tasks.Yep. Though I think they are few and far in between. -- Grant. . . . unix || die smime.p7s Description: S/MIME Cryptographic Signature
Re: free collaborative tools for low BW and losy connections
some of us still do uucp, over tcp and over pots. archaic, but still the right tool for some tasks. randy
Re: free collaborative tools for low BW and losy connections
On Wed, Mar 25, 2020 at 09:59:53AM -0600, Grant Taylor via NANOG wrote: > Something that might make you groan even more than NNTP is UUCP. UUCP > doesn't even have the system-to-system (real time) requirement that NNTP > has. It's quite possible to copy UUCP "Bag" files to removable media and > use sneaker net t transfer things. I was remiss not to mention this as well. *Absolutely* UUCP still has its use cases, sneakernetting data among them. It's been a long time since "Never underestimate the bandwidth of a station wagon full of tapes" (Dr. Warren Jackson, Director, UTCS) but it still holds true for certain values of (transport container, storage medium). ---rsk
Re: free collaborative tools for low BW and losy connections
Thanks, my facepalm moment of the day (so far; it's only 7:30am here) is... Use tools from the past when the connections everywhere were losy and slow. They already mentioned RT. I'll mention that and NNTP/UUCP/etc. scott
Re: free collaborative tools for low BW and losy connections
Paul Ebersman wrote on 25/03/2020 16:59: And scary as it sounds, UUCP over SLIP/PPP worked remarkably robustly. uucp is a batch oriented protocol so it's pretty decent for situations where there's no permanent connectivity, but uncompelling otherwise. nntp is a non-scalable protocol which broke under its own weight. Threaded news-readers are a great way of catching up with large mailing lists if you're prepared to put in the effort to create a bidirectional gateway. But that's really a statement that mail readers are usually terrible at handling large threads rather than a statement about nntp as a useful media delivery protocol. Nick
Re: free collaborative tools for low BW and losy connections
woody> UUCP kicks ass. And scary as it sounds, UUCP over SLIP/PPP worked remarkably robustly. When system/network resources are skinny or scarce, you get really good at keeping things working. :)
Re: free collaborative tools for low BW and losy connections
> On Mar 25, 2020, at 4:59 PM, Grant Taylor via NANOG wrote: > UUCP doesn't even have the system-to-system (real time) requirement that NNTP > has. Brian Buhrow and I replaced a completely failing database-synchronization-over-Microsoft-Exchange system with UUCP across American President Lines and Neptune Orient Lines fleets, back in the mid-90s. UUCP worked perfectly (Exchange connections were failing ~90% of the time), was much faster (average sync time on each change reduced from about three minutes to a few seconds), and saved them several million dollars a year in satellite bandwidth costs. UUCP kicks ass. -Bill signature.asc Description: Message signed with OpenPGP
Re: free collaborative tools for low BW and losy connections
In article <9f22cde2-d0a2-1ea1-89e9-ae65c4d47...@tnetconsulting.net> you write: >I hadn't considered having a per system NNTP server. I sort of like the >idea. I think it could emulate the functionality that I used to get out >of Lotus Notes & Domino with local database replication. I rarely >needed the offline functionality, but having it was nice. I also found >that the local database made searches a lot faster than waiting on them >to traverse the network. > >> Also note that bi- or unidirectional NNTP/SMTP gateways are useful. I've been reading nanog and many other lists on my own NNTP server via a straightforward mail gateway for about a decade. Works great. I'm sending this message as a mail reply to a news article. -- Regards, John Levine, jo...@taugh.com, Primary Perpetrator of "The Internet for Dummies", Please consider the environment before reading this e-mail. https://jl.ly
Re: free collaborative tools for low BW and losy connections
On 3/25/20 5:39 AM, Rich Kulawiec wrote:> One of the tools that we've had for a very long time but which is often overlooked is NNTP. It's an excellent way to move information around under exactly these circumstances: low bandwidth, lossy connections -- and intermittent connectivity, limited resources, etc. I largely agree. Though NNTP does depend on system-to-system TCP/IP connectivity. I say system-to-system instead of end-to-end because there can be intermediate systems between the end systems. NNTP's store and forward networking quite capable. Something that might make you groan even more than NNTP is UUCP. UUCP doesn't even have the system-to-system (real time) requirement that NNTP has. It's quite possible to copy UUCP "Bag" files to removable media and use sneaker net t transfer things. I've heard tell of people configuring UUCP on systems at the office, their notebook that they take with them, and systems at home. The notebook (push or poll) connects to the systems that it can currently communicate with and transfers files. UUCP can also be used to transfer files, news (NNTP: public (Usenet) and / or private), email, and remote command execution. Nearly any laptop/desktop has enough computing capacity to run an NNTP server Agreed. I dare say that anything that has a TCP/IP stack is probably capable of running an NNTP server (and / or UUCP). depending on the quantity of information being moved around, it's not at all out of the question to do exactly that, so that every laptop/desktop (and thus every person) has their own copy right there, thus enabling them to continue using it in the absence of any connectivity. I hadn't considered having a per system NNTP server. I sort of like the idea. I think it could emulate the functionality that I used to get out of Lotus Notes & Domino with local database replication. I rarely needed the offline functionality, but having it was nice. I also found that the local database made searches a lot faster than waiting on them to traverse the network. Also note that bi- or unidirectional NNTP/SMTP gateways are useful. Not only that, but given the inherent one-to-many nature of NNTP, you can probably get away with transmitting that message once instead of (potentially) once per recipient. (Yes, I know that SMTP is supposed to optimize this, but I've seen times when it doesn't work, properly.) It's not fancy, but anybody who demands fancy at a time like this is an idiot. It *works*, it gets the basics done, and thanks to decades of development/experience, it holds up well under duress. I completely agree with your statement about NNTP. I do think that UUCP probably holds up even better. UUCP bag files make it easy to bridge communications across TCP/IP gaps. You could probably even get NNTP and / or UUCP to work across packet radio. }:-) -- Grant. . . . unix || die smime.p7s Description: S/MIME Cryptographic Signature
Re: free collaborative tools for low BW and losy connections
One of the tools that we've had for a very long time but which is often overlooked is NNTP. It's an excellent way to move information around under exactly these circumstances: low bandwidth, lossy connections -- and intermittent connectivity, limited resources, etc. Nearly any laptop/desktop has enough computing capacity to run an NNTP server and depending on the quantity of information being moved around, it's not at all out of the question to do exactly that, so that every laptop/desktop (and thus every person) has their own copy right there, thus enabling them to continue using it in the absence of any connectivity. Also note that bi- or unidirectional NNTP/SMTP gateways are useful. It's not fancy, but anybody who demands fancy at a time like this is an idiot. It *works*, it gets the basics done, and thanks to decades of development/experience, it holds up well under duress. ---rsk
Re: free collaborative tools for low BW and losy connections
It would be a lot MORE relevant if there were some actual tools listed & discussed! Miles Fidelman On 3/24/20 1:48 PM, Scott Weeks wrote: Hello, I was watching SDNOG and saw the below conversation recently. Here is the relavant part: "I think this is the concern of all of us, how to work from home and to keep same productivity level,, we need collaborative tools to engaging the team. I am still searching for tool and apps that are free, tolerance the poor internet speed." I know of some free tools and all, but am not aware of the tolerance they may have to slow speed and (likely) poor internet connections. I was wondering if anyone here has experience with tools that'd work, so I could suggest something to them. I don't know if everyone's aware of what they have been going through in Sudan (both of them), but it has been a rough life there recently. Thanks! scott Original Message Subject: [sdnog] How to work from home Hi all Hope you are all safe wherever you are, Regards to the current situation around the world , and as we all adviced/forced to start working from home which is not common here in our community , and I know some bosses are not convinced unless they saw you in your desk :D my question is , for simple offices ,with no great infrastructure , just an internet connection to their edge ,how can they work from home ? Is there any free tools /ways they can use, what are the options, with taking along the security concerns what is your advice to achieve that in a proper way , and for those who managed to work from home , how did you do that ? Please share your experience ^_^ And how we as "sdnog community" can help in that "for the old fashioned bosses :D" -- From: "aseromeru...@hotmail.com" I think this is the concern of all of us, how to work from home and to keep same productivity level,, we need collaborative tools to engaging the team. I am still searching for tool and apps that are free, tolerance the poor internet speed. Any suggestion -- In theory, there is no difference between theory and practice. In practice, there is. Yogi Berra Theory is when you know everything but nothing works. Practice is when everything works but no one knows why. In our lab, theory and practice are combined: nothing works and no one knows why. ... unknown
