Re: Fed Bill Would Restrict Web Server Logs
On Tue, Feb 14, 2006 at 09:47:50AM -0500, Jon R. Kibler scribed: http://www.politechbot.com/docs/markey.data.deletion.bill.020806.pdf to delete information about visitors, including e-mail addresses, if the data is no longer required for a legitimate business purpose. Original posting from Declan McCullagh's PoliTech mailing list. Thought NANOGers would be interested since, if this bill passes, it would impact almost all of us. Just imagine the impact on security of not being able to login IP address and referring page of all web server connections! Call me weird, but I fail to see where the scary teeth lie in such a bill. First of all, it's phrased very abstractly and would hopefully have its language clarified by the time it escapes a committee. Second, the bill is fairly clear about the meaning of personal information, and it doesn't include things like IP addresses in its examples; the latter would be a matter for a court to decide, and it's not clear cut at all: ... that allows a living person to be identified individually, including ... : first and last name, home or physical address, ... Third, it says nothing at all about restricting what you can log: An owner of an Internet website shall destroy, within a reasonable period of time, any data containing personal information if the information is no longer necessary for the purpose for which it was collected or any other legitimate business purpose. If you need IP address logging to ensure the security of your website, then that sounds like a pretty legitimate business practice. The more interesting question is how _long_ you need to keep the personal information around for your for your legitimate business purposes. A week? A month? A year? Ultimately, it would probably boil down to a dash of best practices and a pinch of CYA. But there's nothing in there to freak out about for day to day operations. The worry is more that you'd probably have to ensure that your logs get blasted or sanitized according to a well-defined schedule. Which, when you think about it, might not be a bad thing at all. -Dave -- Dave Andersen [EMAIL PROTECTED] Assistant Professor 412.268.3064 Carnegie Mellon Universityhttp://www.cs.cmu.edu/~dga
Re: multi homing pressure
On Wed, Oct 19, 2005 at 10:19:28PM +, Paul Vixie scribed: [EMAIL PROTECTED] (Jared Mauch) writes: it will be interesting to see if this has acutal impact on ASN allocation rates globally. i don't think so. multihoming without bgp isn't as hard as qualifying for PI space. i think we'll finally see enterprise-sized multihoming NAT/proxy products. If you can run Squid, you can multihome your web connections today. It's a little bit awkward to configure, but then again, so is Squid. People are welcome to poke at, fold, spindle, or mutilate: http://nms.lcs.mit.edu/ron/ronweb/#code (Part of my thesis work, Monet is a modification to Squid that causes it to try to open N TCP connections to a Web server that it wants to talk to. It uses the first SYN ACK to return, and closes the other connections to be a nice neighbor. It's shockingly effective at improving availability to Web sites that are themselves multihomed or otherwise good. Warning: Often still leads to annoyance if you find yourself able to browse the web but not do anything else. We do have a NAT version of this that works with arbitrary protocols. If people are interested, I'll try to convince my former student to dig up the code and make it a bit prettier.) -Dave
Re: multi homing pressure
On Thu, Oct 20, 2005 at 03:18:35AM +0100, Paul Jakma scribed: On Wed, 19 Oct 2005, David G. Andersen wrote: If you can run Squid, you can multihome your web connections today. It's a little bit awkward to configure, but then again, so is Squid. People are welcome to poke at, fold, spindle, or mutilate: http://nms.lcs.mit.edu/ron/ronweb/#code (Part of my thesis work, Hehe, google for vixie ifdefault. Right. Vix was talking about the inbound path - I'm talking about the outbound path. Complimentary solutions to the same problem. -Dave
Sites wanted for research boxes
I sent a similar mail out a couple of years ago and greatly appreciate the response I got. Time and entropy have done their dirty work, so we're looking for a few (more) good hosts. We've been running a moderate sized (30 node) overlay network and general network research testbed for the last 4 or 5 years. The testbed started as part of the Resilient Overlay Networks project at MIT, and has evolved to support a variety of research by about 15 network researchers at a number of institutions. At this time, the primary institutions involved in the testbed management are MIT, CMU, NYU, and the University of Utah. (Regular NANOG attendees may know Nick Feamster from some talks he's given about routing and automated tools for managing and debugging sets of router configurations. Nick's the MIT contact for the testbed these days.) Our major goal for the testbed is to have access to a realistic set of Internet paths - we want to make sure that network research actually takes place in the real world, not a perfect isolated environment. Which is where nanog comes in... What we need is a machine we can place on your network. We will be happy to give you the machine (a PC), which you can use as well for your own work. Ideally, it is best if you can place it outside your firewall (if you have one), since that may simplify logistics at your end. This PC will run a FreeBSD kernel provided by us; all you need to do is to configure (or DHCP) an IP address, and we'll be all set. The machine will have a 10/100 Ethernet interface. We have two options for bandwidth usage. If you'll let us, we'd love to be able to consume a fair bit of it from time to time. Some of our researchers are experimenting with data transfer protocols and more efficient ways of shipping things (like news) around the 'net. However, if you're bandwidth constrained, we do a lot of measurements and other low-bandwidth experiments that benefit greatly from just having a fairly non-intrusive presense. We'd love a BGP feed, or even internal routing feeds, for data collection. We won't be sniffing packets on the network, etc. and we will work to ensure that the machine is as secure as we can make it - most services will be disabled, and the running services will be firewalled, and we'll keep the machine up to date with security patches. (No problems yet, fingers crossed...) What you get in return: a) A locally hosted stratum 1 time source that you're welcome to use or let your clients use. Our machines are CDMA synchronized. b) Our eternal gratitude, love, and acknowledement (and participation / input, if you are interested!) in our research. This also involves free beer if you swing by Pittsburgh. :) c) One of our goals is to create tools that run on the testbed that are useful to their hosts for things like distributed debugging, or the aforementioned BGP configuration debugging tools. We also collect BGP feeds and have a nice interface for searching through the historical data for figuring out what went wrong on the 'net at a particular time. d) Ask! If you have specific network problems/etc., we're always looking for more problems to solve. Your problems are our dinners; we'd like to know what they are and try to create solutions. Please let us know if you can help; we'd appreciate it very much! If so, could you please tell us if: 1. You prefer a small (1U) rack-mountable machine for us to send you as a package, OR 2. If you have, and are willing to, spare your own PC, we can give you a disk image that you can burn, insert, enter an IP address, and be good to go. -Dave -- work: dga at cs.cmu.edu du me: [EMAIL PROTECTED] Carnegie Mellon Universityhttp://www.angio.net/ Department of Computer Science
Re: BCP38 making it work, solving problems
On Tue, Oct 19, 2004 at 07:14:32PM +0200, JP Velders scribed: Date: Tue, 19 Oct 2004 09:21:46 -0700 From: Randy Bush [EMAIL PROTECTED] Subject: Re: BCP38 making it work, solving problems For example, how many ISPs use TCP MD5 to limit the possibility of a BGP/TCP connection getting hijacked or disrupted by a ddos attack? i hope none use it for the latter, as it will not help. more and more use it for the former. why? becuase they perceived the need to solve an immediate problem, a weakness in a vendor's code. Uhm, you might need to run that by me again... Hijacking the connection is in a completely different class as someone bombarding you with a bunch of forged BGP packets to close down a session. Without that MD5 checksum you are quite vulnerable to that. I haven't seen a vendor come up with a solution to that, because the problem is on a much more vendor-neutral level... Unless you're worried about an adversary who taps into your fiber, how is MD5 checksums any better than anti spoofing filters that protect your BGP peering sessions? The only benefit I see is that you can actually verify that your peer is using md5 checksums, instead of having to take them on faith that they won't permit someone to spoof their router's address. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/
Re: That MIT paper
On Thu, Aug 12, 2004 at 01:35:36PM +0200, Niels Bakker scribed: * [EMAIL PROTECTED] (David G. Andersen) [Thu 12 Aug 2004, 02:55 CEST]: Global impact is greatest when the resulting load changes are concentrated in one place. The most clear example of that is changes that impact the root servers. When a 1% increase in total traffic is instead spread among hundreds of thousands of different, relatively unloaded DNS servers, the impact on any one DNS server is minimal. And since we're talking about a protocol that variously occupies less than 3% of all Internet traffic, the packet count / byte count impact is negligible (unless it's concentrated, as happens at root and gtld servers). This doesn't make sense to me. You're saying here that a 1% increase in average traffic is a 1% average increase in traffic. What's your point? if a load change is concentrated in one place how can the impact be global? Because that point could be critical infrastructure (to abuse the buzzword). If a 1% increase in DNS traffic is 100,000 requests per second (this number is not indicative of anything, just an illustration), that could represent an extra request per second per nameserver -- or 7,000 more requests per second at the root. One of these is pretty trivial, and the other could be unpleasant. At root and gTLD servers I assume DNS traffic occupies significantly more than 3% of all traffic there. Still, a 1% increase remains 1%. Sure, but the ratio still plays out. If your total traffic due to DNS is small, then even a large (percentage) increase in DNS traffic doesn't affect your overall traffic volume, though it might hurt your nameservers. If you're a root server, doubling the DNS traffic nearly doubles total traffic volume, so in addition to DNS-specific issues, you'll also start looking at full pipes. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/
Re: That MIT paper
On Wed, Aug 11, 2004 at 04:49:18PM +, Paul Vixie scribed: what i meant by act globally, think locally in connection with That MIT Paper is that the caching effects seen at mit are at best representative of that part of mit's campus for that week, and that Totally agreed. The paper was based upon two traces, one from MIT LCS, and one from KAIST in Korea. I think that the authors understood that they were only looking at two sites, but their numbers have a very interesting story to tell -- and I think that they're actually fairly generalizable. For instance, the rather poorly-behaving example from your f-root snapshot is rather consistent with one of the findings in the paper: [Regarding root and gTLD server lookups] ...It is likely that many of these are automatically generated by incorrectly implemented or configured resolvers; for example, the most common error 'loopback' is unlikely to be entered by a user even a variance of 1% in caching effectiveness at MIT that's due to generally high or low TTL's (on A, or MX, or any other kind of data) becomes a huge factor in f-root's load, since MIT's load is only one But remember - the only TTLs that the paper was suggesting could be reduced were non-nameserver A records. You could drop those all to zero and not affect f-root's load one bit. In fairness, I think this is jumbled together with NS record caching in the paper, since most responses from the root/gTLD servers include both NS records and A records in an additional section. Global impact is greatest when the resulting load changes are concentrated in one place. The most clear example of that is changes that impact the root servers. When a 1% increase in total traffic is instead spread among hundreds of thousands of different, relatively unloaded DNS servers, the impact on any one DNS server is minimal. And since we're talking about a protocol that variously occupies less than 3% of all Internet traffic, the packet count / byte count impact is negligible (unless it's concentrated, as happens at root and gtld servers). The other questions you raise, such as: how much of the measured traffic was due to bad logic in caching/forwarding servers, or in clients? how will high and low ttl's affect bad logic that's known to be in wide deployment? are equally important questions to ask, but .. there are only so many questions that a single paper can answer. This one provides valuable insight into client behavior and when and why DNS caching is effective. There have been other papers in the past (for instance, Danzig's 1992 study) that examined questions closer to those you pose. The results from those papers were useful in an entirely different way (namely, that almost all root server traffic was totally bogus because of client errors). It's clear that from the perspective of a root name server operator, the latter questions are probably more important. But from the perspective of, say, an Akamai or a Yahoo (or joe-random dot com), the former insights are equally valuable. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/
Re: that MIT paper again
Regarding both Paul's message below and Simon Walter's earlier message on this topic... Simon Walters scribed: I'm slightly concerned that the authors think web traffic is the big source of DNS, they may well be right (especially given one of the authors is talking about his own network), but my quick glance at the Two things - first, the paper breaks down the DNS traffic by the protocol that generated it - see section III C, which notes a small percentage of these lookups are related to reverse bloack-lists such as rbl.maps.vix.com -- but remember that the study was published in 2001 based upon measurements made in January and December of 2000. RBL traffic wasn't nearly the proportion of DNS queries that it is today. As the person responsible for our group's spam filtering (one mailserver among many that were measured as a part of the study), we didn't start using spamassassin until late 2001, and I believe we were one of the more aggressive spam filtering groups in our lab. Also note that they found that about 20% of the TCP connections were FTP connections, mostly to/from mirror sites hosted in our lab. Sendmail of five years ago also wasn't as aggressive about performing reverse verification of sender addresses. I asked Jaeyeon about this (we share an office), and she noted that: In our follow-up measurement study, [we found] that DNSBL related DNS lookups at CSAIL in February 2004 account for 14% of all DNS lookups. In comparison, DNSBL related traffic accounted for merely 0.4% of all DNS lookups at CSAIL in December 2000. Your question was right on the money for contemporary DNS data. The abstract doesn't mention that the TTL on NS records is found to be important for scalability of the DNS. Probably the main point Paul wants us to note. Just because the DNS in insensitive to slight changes in A record TTL doesn't mean TTL doesn't matter on other records. This is a key observation, and seems like it's definitely missing from the abstract (alas, space constraints...). They're not talking about the NS records, and they're not talking about the associated A records for _nameservers_. On Sat, Aug 07, 2004 at 04:55:00PM +, Paul Vixie scribed: here's what i've learned by watching nanog's reaction to this paper, and by re-reading the paper itself. 1. almost nobody has time to invest in reading this kind of paper. 2. almost everybody is willing to form a strong opinion regardless of that. 3. people from #2 use the paper they didn't read in #1 to justify an opinion. :) human nature. 4. folks who need academic credit will write strong self-consistent papers. 5. those papers do not have to be inclusive or objective to get published. 6. on the internet, many folks by nature think locally and act globally. 7. #6 includes manufacturers, operators, endusers, spammers, and researchers. 8. the confluence of bad science and disinterested operators is disheartening. 9. good actual policy must often fly in the face of accepted mantra. I'm not quite sure how to respond to this part (because I'm not quite sure what you meant...). It's possible that the data analyzed in the paper may not be representative of, say, commercial Internet traffic, but how is the objectivity in question? The conclusions of the paper are actually pretty consistent with what informed intuition might suggest. First: If NS records had lower TTL values, essentially all of the DNS lookup traffic observed in our trace would have gone to a root or gTLLD server, which would have increased the load on them by a factor of about five. Good NS-record caching is therefore critical to DNS scalability. and second: Most of the benefit of caching [of A records] is achieved with TTL values of only a small number of minutes. This is because most cache hits are produced by single clients looking up the same server multiple times in quick succession [...] As most operational experience can confirm, operating a nameserver for joe-random-domain is utterly trivial -- we used to (primary) a couple thousand domains on a p90 with bind 4.. As your own experience can confirm, running a root nameserver is considerably less trivial. The paper confirms the need for good TTL and caching management to reduce the load on root nameservers, but once you're outside that sphere of ~100 critical servers, the hugely distributed and heavy-tailed nature of DNS lookups renders caching a bit less effective except in those cases where client access patterns cause intense temporal correlations. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/
L3 burp today - what happened?
Anyone know what happened to L3 during the last hour? They seem to have developed an appetite for dropping packets in San Jose for customers on the Genuity portion of their network, but I'm curious if anyone has a slightly more detailed explanation about the failure. The failure seems to have started at 17:09 and ended at about 17:51 EST. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/
Re: Does anyone think it was a good thing ? (was Re: VeriSign Capitulates
On Fri, Oct 03, 2003 at 05:34:05PM -0400, jeffrey.arnold quacked: On Fri, 3 Oct 2003, Mike Tancsa wrote: :: OK, so was ANYONE on NANOG happy with :: a) Verisign's site finder :: b) How they launched it :: Disregarding their implementation issues, the product is pretty good. I've actually used it to fix a few typos, etc... From an end user perspective, it's certainly better than a squid error page. Yeah, but this is easy for you to provide as a service to users who want it. patch your squids with the following change to src/errorpage.c: @@ -414,6 +414,7 @@ * T - UTC x * U - URL without password x * u - URL with passwordx + * V - URL without http method without password x * w - cachemgr email address x * z - dns server error message x */ @@ -546,6 +547,9 @@ case 'u': p = r ? urlCanonical(r) : err-url ? err-url : [no URL]; break; +case 'V': + p = r ? urlCanonicalStripped(r) : err-url ? err-url : [no URL]; +break; case 'w': if (Config.adminEmail) memBufPrintf(mb, %s, Config.adminEmail); And then modify errors/English/ERR_DNS_FAIL to say: H2Alternatives/H2 You can try to view this server through: ul li a href=http://www.google.com/search?q=cache:%V;Google's Cache/a li a href=http://web.archive.org/web/*/%U;The Internet Archive/a li a href=http://sitefinder.verisign.com/lpc?url=%V;Use Sitefinder to search for typos for this domain/a /ul If you're creative, have it send them with a redirect to a local CGI script that tries obvious typos. Very simple. My users like the link to the internet archive (also modify the could not connect error page and others). If you just want HTML, create a framed document that auto-loads the sitefinder doc in the bottom half, and pops up your own error page in the front. I leave that as an exercise to the HTML-clued reader, but it's not very hard. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: anycast (Re: .ORG problems this evening)
On Thu, Sep 18, 2003 at 02:38:18PM -0400, Todd Vierling quacked: On Thu, 18 Sep 2003, E.B. Dreger wrote: : EBD That's why one uses a daemon with main loop including : EBD something like: : EBD : EBDsuccess = 1 ; : EBDfor ( i = checklist ; i-callback != NULL ; i++ ) : EBDsuccess = i-callback(foo) ; : EBDif ( success ) : EBDsend_keepalive(via_some_ipc_mechanism) ; Yes, I hope that UltraDNS implements something like this, if they have not already. It's still not a guarantee that things will get withdrawn -- or be reachable, even if working but not withdrawn -- in case of a problem. That still leaves the DNS for a gTLD at risk for a single point of failure. The whole problem with only listing two anycast servers is that you leave yourself vulnerable to other kinds of faults. Your upstream ISP fat-fingers ip route 64.94.110.11 null0 and accidentally blitzes the netblock from which the anycast servers are announced. A router somewhere between customers and the anycast servers stops forwarding traffic, or starts corrupting transit data, without interrupting its route processing. packet filters get misconfigured.. (Observe how divorced route processing and packet processing are in modern routing architectures and it's pretty easy to see how this can happen. With load balancing, traffic can get routed down a non-functional path while routing takes place over the other one - BBN did that to us once, was very entertaining). Route updates in BGP take a while to propagate. Much longer than the 15ms RTT from me to, say, a.root-server.net. The application retry in this context can be massively faster than waiting 30+ seconds for a BGP update interval. The availability of the DNS is now co-mingled with the success of the magic route tweak code; the resulting system is a fair bit more complex than simply running a bunch of different DNS servers. God forbid that zebra ever has bugs... http://www.geocrawler.com/lists/3/GNU/372/0/ In contrast, talking to a few DNS servers gives you an end-to-end test of how well the service is working. You still depend on the answers being correct, but you can intuit a lot from whether or not you actually get answers, instead of sitting around twiddling your thumbs thinking, gee, I sure wish that routing update would get sent out so I could use the 'net. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: News of ISC Developing BIND Patch
On Wed, Sep 17, 2003 at 02:50:51AM -0700, Vadim Antonov quacked: In fact, we do have an enormously useful and popular way of doing exactly that - this is called search engines and bookmarks. What is needed is an infrastructure for allocation of unique semantic-free end point identifiers (to a large extent, MAC addresses may play this role, or, say, 128-bit random numbers), a way to translate EIDs to the topologically allocated IP addresses (a kind of simplified numbers-only DNS?) and a coordinated effort to change applications and expunge domain names from protocols, databases, webpages and such, replacing URLs containing domain names with URLs containing EIDs. Oh, you mean something like the Semantic Free Referencing project? http://nms.lcs.mit.edu/projects/sfr/ (Blatant plug for a friend's research, yes, but oh my god does it seem relevant today) -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: Private port numbers?
On Wed, Aug 13, 2003 at 10:40:30PM +, Christopher L. Morrow quacked: what about ports that start as 'private' and are eventually ubiquitously used on a public network? (Sean Donelan noted that 137-139 were originally intended to be used in private networks... and they became 'public' over time) You run it on a different port. I actually really like this idea, because it makes shipping a more secure default configuration easier for vendors without having to coordinate between firewall vendors and implementors. The gotcha is that it makes life pretty weird for you if you then want to make your service work in the wide-area... but that's pretty easy to do with intelligent defaults: Ports 1-1024: Well-known-ports Ports 60001-61024: Private well-known-port analogues Applications would try: if (!connect(..., public port #)) connect(..., private port #)) In fact, this (impractically) generalizes to a nice way of signifying whether or not you want external people to be able to talk to your service: port bit[0] == 0: Public service, please do not filter port bit[0] == 1: Private service, please filter at organizational boundary I suddenly wish the port space was 32 bits. :) People _could_, of course, implement all of this with tcpwrappers and host-local firewalls. But experience has shown that they don't. It might be easier for them if they could just click private when they configured the service, though experience has shown that services migrate to the less restrictive mode as debugging and time goes on... -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Complaint of the week: Ebay abuse mail (slightly OT)
To add to the eternally annoying list of companies that ignore abuse@ mail... ebay now requires that you fill in their lovely little web form to send them a note. Even if, say, you're trying to let them know about another scam going around that tries to use the machine www.hnstech.co.kr to extract people's credit card information. Has anyone had success in convincing companies that this is just A Bad Idea (ignoring abuse mail), and if so, how did you manage to do it? Sorry for the slightly non-operational content, but I've had it with ebay on this one. -Dave - Forwarded message from eBay Safe Harbor [EMAIL PROTECTED] - Date: Sat, 02 Aug 2003 22:58:01 -0700 From: eBay Safe Harbor [EMAIL PROTECTED] Subject: Your message to [EMAIL PROTECTED] was not received (KMM86277800V90276L0KM) To: David G. Andersen [EMAIL PROTECTED] Auto-Submitted: auto-replied Reply-To: eBay Safe Harbor [EMAIL PROTECTED] X-MIME-Autoconverted: from quoted-printable to 8bit by eep.lcs.mit.edu id h735w5sU087612 Thank you for writing to the eBay SafeHarbor Team. The address you wrote to ([EMAIL PROTECTED]) is no longer in service. Please resend your email to us through one of the online webforms listed below. Using these forms will help us direct your email to the right department where we can quickly answer your question correctly and get it right back to you. For Trust and Safety issues (reports of policy violations, problems with transactions, suspensions, etc.) please use the following webform: http://pages.ebay.com/help/basics/select-RS.html For General Support questions (billing, bidding, or selling concerns and technical issues, etc.) please use the following webform: http://pages.ebay.com/help/basics/select-support.html Once on the webform, what will really help us assist you further is if you choose the best topic for your question. This will allow you to view our Instant Help pages, where you may find your answer immediately. Should you not find your answer there, choosing the best topics will still help us answer your question faster, correctly, and completely. We truly appreciate your assistance in this matter and apologize for any inconvenience this may have caused. Sincerely, eBay SafeHarbor Team - End forwarded message - -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: North America not interested in IP V6
On Thu, Jul 31, 2003 at 11:02:14AM -0600, Irwin Lazar quacked: As one person noted in response to Christian's speech. If there is no addressing shortage, why do I have to pay $75 a month for a DSL connection with a static IP address when a floating IP address only costs me $40 per month? I think there are two parts to the answer. a) DHCP'ing everyone is just easier. b) Why do you pay less for a flight with a saturday night stopover? - Market segmentation. People with static addresses usually want to do things like run servers, and are probably willing to pay for the privilege. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: Latency generator?
On Wed, Jun 25, 2003 at 12:48:29PM -0400, Temkin, David quacked: Does anyone know of any free, cheap, or potentially rentable latency generators? Ideally I'd like something that just sits between two ethernet devices to induce layer 2/3 latency in traffic, but am open to any options... Dummynet. We use it at Emulab (http://www.emulab.net/) to do exactly what you're describing. You have to use it in conjunction with the bridging code, and then you can just do it. By default, it still uses the ipfw firewall rules to match traffic, so it only delays IP, but that could probably be fixed with a little hacking if you also want to delay ARP and other things. Built into FreeBSD. Should work mostly out of the box. It'll also do traffic shaping and whatnot. Your .signature disclaimer was longer than your message, by the way. ;-) -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: NAT for an ISP
On Wed, Jun 04, 2003 at 12:51:51PM -0700, Christopher J. Wolff quacked: Hello, I would like to know if any service providers have built their access networks out using private IP space. It certainly would benefit the global IP pool but it may adversely affect users with special applications. At any rate, it sounds like good fodder for a debate. I've got a friend who puts all of his internal servers, routers, and _customers_ on RFC1918 space and pipes them out thrugh a PNAT. Fairly small ISP - maybe 15 megabits of bandwidth - operating at the state local level. It's an interesting setup. Kind of fun. The stateful pnat functionality forces customers to specify exactly what inbound services they want, which can't hurt security. Every customer gets a /24 or greater, which helps convenience. On the other hand, everyone has a NAT in front of them, which means that they get clients who would have probably been putting a NAT in front of themselves anyway. I probably wouldn't use that setup myself, but then again, I subscribe to nanog... -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: NAT for an ISP
On Wed, Jun 04, 2003 at 07:07:28PM -0400, Andy Dills quacked: I've got a friend who puts all of his internal servers, routers, and _customers_ on RFC1918 space and pipes them out thrugh a PNAT. Fairly small ISP - maybe 15 megabits of bandwidth - operating at the state local level. Why on earth would they do this? What you've said implies DS3 level connectivity, so to skimp on ARIN fees seems a little ridiculous. Historical accident in many ways. I implied DS3-level connectivity, but what it really means is multiple bonded T1s from multiple providers. It started out as a T1 from here, a T1 from there, and no local BGP knowledge (and discouragement from the upstreams). In fact, using a bunch of NATs is a great way to resell cheap upstream connections. Yeah, I read you loud and clear. My friend is a half-baked cluebie using techniques I'll term fun and later encourage my competitors to employ. :) Actually, I do mean the fun part. You can do some cool tricks with it. Renumbering to different providers is mostly seamless, particularly since he runs the DNS for his customers. Easy to experiment with throwing transparent caches and things like that in front of the customers since they're already going through a firewall. Now that he's about large enough to get ARIN space, the game is changing, and they're moving in the directions one would expect them to. It's not an approach that I would ever encourage a large ISP to take. In fact, I don't necessarily think of him as providing standard Internet services - he provides primarily web, mail, and VPN services, and then some customized stuff on a per-customer basis. But he's had a decent customer base for a small ISP, and he seems to be filling a niche, and hasn't gone out of business doing it. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: Non-GPS derived timing sources (was Re: NTp sources that work in a datacenter)
On Sun, Jun 01, 2003 at 08:13:08AM -0700, Peter Lothberg quacked: I don't expect GPS to spin out of control soon.. So GPS tracks TAI and the difference is published (2 months after the fact..) But it's simple to build a 'jamer' that makes GPS reception not work in a limited area, same for Loran-C used in combination with GPS in many Sonet/SDH S1 devices. but I did wonder how hard it is to find a another reliable clock source of similar quality to GPS to double check GPS. For NTP purposes, WWVB is actually just fine, as long as you properly configure your distance from the transmitter. The NTP servers list shows several WWVB synchronized clocks. CDMA clocks synch indirectly to GPS, but are typically locally stabalized by a rubidium or ovenized quartz oscillator with decent holdover capabilities for a few days of GPS outages. But they'll suffer the same fate if GPS went just plain wrong. The NIST timeservers are available over the net, if you can deal with that degree of synch. Lots of them just use ACTS dialup synch to get the offset, and have very good local clocks. ACTS is certainly a good fall-back for GPS, since it uses a wired path instead of a wireless one. So if you're really paranoid: GPS + WWVB + ACTS + internet to tick/tock or one of the NIST primaries. Ultimately, WWVB, ACTS, and ntp to NIST are all synched from pretty much the same source, but the odds that they'd all go bad are pretty slim. GPS is steered from the USNO time, but the clocks on the satellites are pretty good. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: 923Mbits/s across the ocean
On Sun, Mar 09, 2003 at 02:25:25PM +0100, Iljitsch van Beijnum quacked: On Sat, 8 Mar 2003, Joe St Sauver wrote: you will see that for bulk TCP flows, the median throughput is still only 2.3Mbps. 95th%-ile is only ~9Mbps. That's really not all that great, throughput wise, IMHO. Strange. Why is that? RFC 1323 is widely implemented, although not widely enabled (and for good reason: the timestamp option kills header compression so it's bad for lower-bandwidth connections). My guess is that the OS can't afford to throw around MB+ size buffers for every TCP session so the default buffers (which limit the windows that can be used) are relatively small and application programmers don't override the default. Which makes it doubly a shame that the adaptive buffer tuning tricks haven't made it into production systems yet. It was a beautiful, simple idea that worked very well for adapting to long fat networks: http://www.acm.org/sigcomm/sigcomm98/tp/abs_26.html -dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: 923Mbits/s across the ocean
On Sat, Mar 08, 2003 at 03:29:56PM -0500, [EMAIL PROTECTED] quacked: High speeds are not important. High speeds at a *reasonable* cost are important. What you are describing is a high speed at an *unreasonable* cost. To paraphrase many a california sufer, dude, chill out. The bleeding edge of performance in computers and networks is always stupidly expensive. But once you've achieved it, the things you did to get there start to percolate back into the consumer stream, and within a few years, the previous bleeding edge is available in the current O(cheap) hardware. A cisco 7000 used to provide the latest and greatest performance in its day, for a rather considerable cost. Today, you can get a box from Juniper for the same price you paid for your 7000 that provides a few orders of magnitude more performance. But to get there, you have to be willing to see what happens when you push the envelope. That's the point of the LSR, and a lot of other research efforts. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: 923 Mbps across the Ocean ...
On Fri, Mar 07, 2003 at 10:09:51PM +0100, Mikael Abrahamsson quacked: On Fri, 7 Mar 2003, Richard A Steenbergen wrote: Production commercial networks need not apply, 'lest someone realize that they blow away these speed records on a regular basis. What kind of production environment needs a single TCP stream of data at 1 gigabit/s over a 150ms latency link? Just the fact that you need a ~20 megabyte TCP window size to achieve this (feel free to correct me if I'm wrong here) seems kind of unusal to me. It's unusual, but it's not completely unheard of. One of the biggest sources of such data is VLBI (interferometry to measure the movement of the earth's crust), in which signals from geographically distributed measurement sites have to be recorded and correlated at a central site: http://web.haystack.edu/vlbi/vlbisystems.html The signals are massive. Right now they use specially made tape drives that can record 1Gb/s: ftp://web.haystack.edu/pub/mark4/memos/230.2.pdf ftp://web.haystack.edu/pub/mark4/memos/HDR_concept.PDF and they send the data around via airplanes. They'd love to be able to do real-time correlation of the data, but that involves collecting 6 of these feeds at a central site (more coming). The feeds must be capable of running unattended for up to 24 hours (86 terabytes each, or an aggregate of half a petabyte per day). Yes, backbones push more than a gigabit across links, but not as for a single flow of data. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: anti-spam vs network abuse
On Fri, Feb 28, 2003 at 03:11:00PM -0600, Jack Bates quacked: Should we outlaw a potentially beneficial practice due to its abuse by criminals? Okay. What happens if you make a mistake and overload one of my devices costing my company money. I guarantee you, the law will look favorably on damages. That is the problem with probing. Sometimes the probe itself can be the damage. Programmers are human. Humans make mistakes. Programmers are perfect. That wasn't the question. There are plenty of circumstances in which it's legal to do something once -- say, make a phone call to you and ask how you're doing -- and illegal to do it one hundred million times. You don't outlaw telephones because people can and have used them to harass other people, you outlaw the harassing behavior and make it subject to damages. ... which is exactly what you described. Probing can be knocking on your door, or it can be taking a sledgehammer to your garage. These are so quantitatively different that there is a qualitative shift between the behaviors. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: scripts to map IP to AS?
On Thu, Feb 20, 2003 at 08:09:31AM -0500, William Allen Simpson quacked: Anybody have a pointer to scripts to map IP to AS? There are still 10K-20K hosts spewing M$SQL slammer/sapphire packets, and I'd like to start blocking routing to those irresponsible AS's that haven't blocked their miscreant customers. http://isc.sans.org/port_details.html?port=1434 You can use a quick perl wrapper around whois, or you could use this terribly ugly hacked up traceroute-ng that I wrote to do lookups: http://nms.lcs.mit.edu/software/ron/lookup_as.c Compile with gcc -DSTANDALONE=1 lookup_as.c -o lookup_as -lm And then run. It gets the job done, but it's ugly. :) -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: scripts to map IP to AS?
I should have been a bit more specific. The hacked up traceroute-ng queries the radb, not a whoisd. I've never had problems being blocked when doing radb queries, but YMMV, of course. I also suggest that people be nice and rate-limit their queries so that others don't have to do it for them... -Dave On Thu, Feb 20, 2003 at 12:04:51PM -0500, George Bakos quacked: Careful. Many whoisds don't appreciate automated queries will block YOUR ip address for sometime if you cross their max query rate threshold. You can use a quick perl wrapper around whois, or you could use this terribly ugly hacked up traceroute-ng that I wrote to do lookups: http://nms.lcs.mit.edu/software/ron/lookup_as.c Compile with gcc -DSTANDALONE=1 lookup_as.c -o lookup_as -lm And then run. It gets the job done, but it's ugly. :) -Dave -- George Bakos Institute for Security Technology Studies Dartmouth College [EMAIL PROTECTED] voice 603-646-0665 fax 603-646-0666 Key fingerprint = D646 8F91 F795 27EC FF8B 8C95 B102 9EB2 081E CB85 -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: Level3 routing issues?
On Sun, Jan 26, 2003 at 12:17:20AM -0500, Tim Griffin mooed: hc wrote: I am on Verizon-GNI via Qwest and Genuity and seeing the same problem as well. here's a plot showing the impact on BGP routing tables from seven ISPs (plotted using route-views data): http://www.research.att.com/~griffin/bgp_monitor/sql_worm.html And as an interesting counterpoint to this, this graph shows the number of BGP routing updates received at MIT before, during, and after the worm (3 day window). Tim's plots showed that the number of actual routes at the routers he watched was down significantly - these plots show that the actual BGP traffic was up quite a bit. Probably the withdrawals that were taking routes away from routeviews... http://nms.lcs.mit.edu/~dga/sqlworm.html -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: Level3 routing issues?
On Mon, Jan 27, 2003 at 06:15:33PM -0800, Randy Bush mooed: Wow, for a minute I thought I was looking at one of our old plots, except for the fact that the x-axis says January 2003 and not September 2001 :) :) seeing that the etiology and effects of the two events were quite different, perhaps eyeglasses which make them look the same are not as useful as we might wish? Actually, an eyeballing of the MIT data would suggest that the SQL worm hit harder and faster than NIMDA, and resulted in a more drastic effect on routing tables. I've updated the page I mentioned before: http://nms.lcs.mit.edu/~dga/sqlworm.html to also contain the graph of MIT updates during the NIMDA worm. I should note that our route monitor moved closer to MIT's border router between these updates - it's now colocated in the same datacenter, and before it was across the street, which made it a bit more susceptable to link resets during the NIMDA worm attack. LCS is more prone to dropping off the network than is the entire MIT campus. Therefore, the NIMDA graph probably has a few more session resets (the spikes up to 100,000 routes updated) than it should. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: New worm / port 1434?'
On Sat, Jan 25, 2003 at 10:49:01AM -0500, Eric Gauthier mooed: Ok, I'm not sure if this helps at all. Our campus has two primary connections - the main Internet and something called Internet2. Internet2 has a routing table of order 10,000 routes and includes most top-tier research instituations in the US (and a few other places). By 1am this morning (Eastern US time), all of our Internet links saturated outbound but we didn't appear to see any noticable increase in our Internet2 bandwidth. I'm throwing this out there because it may indicate that the destinations for the traffic - though large - aren't completely random. Has anyone else seen this? It's actually fairly rational. If you look at the size of the I2 routing table in terms of how much of the IP space it covers, it's a fair bit smaller than the full Internet routing table. And most institutions have _more_ I2 bandwidth than commodity internet connectivity. If the probing's roughly random, you'd expect the I2 connection to fare better. MIT's I2 connectivity was better off than its commercial Internet connection as well. Our private peering link to ATT/mediaone was actually in great shape (DS3, very small address space). -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: can you ping mount everest?
On Wed, Jan 22, 2003 at 11:36:14PM -0800, Mike Lyon mooed: The link wants you to log in with a New York Times login... -Mike You can always learn from other mailing lists. username: cipherpunks3 password: cipherpunks -Dave
Re: Is there a line of defense against Distributed Reflective attacks?
On Mon, Jan 20, 2003 at 12:25:27AM -0500, Deepak Jain mooed: As long as the car _moves_ under its own power across the highway, its essentially not the car manufacturers' (or the consumers') immediate concern. That's really not true. Before car companies sell cars, they pass (lots of) safety certification tests. Before owners drive cars legally, they pass a safety and emissions test. Sure, the highway folks clean up after the occasional tire blowout, but there's been a lot of work put in to make sure that the engines aren't going to drop out on a regular basis. If the Internet was a highway, it would be covered in burned-out engines. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: Is there a line of defense against Distributed Reflective attacks?
On Fri, Jan 17, 2003 at 06:38:08PM +, Christopher L. Morrow mooed: has something called Source Path Isolation Engine (SPIE). There This would be cool to see a design/whitepaper for.. Kelly? The long version of the SPIE paper is at: http://nms.lcs.mit.edu/~snoeren/papers/spie-ton.html The two second summary that I'll probably botch: SPIE keeps a (very tiny) hash of each packet that the router sees. If you get an attack packet, you can hand it to the router and ask From where did this come? And then do so to the next router, and so on. The beauty of the scheme is that you can use it to trace single-packet DoS or security attacks as well as flooding attacks. The downside is that it's hardware. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: Is there a line of defense against Distributed Reflective attacks?
On Thu, Jan 16, 2003 at 08:48:03PM -0500, Brad Laue mooed: By way of quick review, such an attack is carried out by forging the source address of the target host and sending large quantities of packets toward a high-bandwidth middleman or several such. One method that comes to mind that can slow the incoming traffic in a more distributed way is ECN (explicit congestion notification), but it doesn't seem as though the implementation of ECN is a priority for many No. ECN is, first and foremost, an optimization for TCP so that it doesn't have to drop packets before cutting its rate back when there's congestion in the network. A zombie or malicious host would just ignore the ECN bit - and the attacks you're describing never reach the point where a host's flow control is involved. You might be thinking of source quench, but that's really not an option with today's networks. Some other conventional alternatives have been discussed already (ingress/egress filtering, etc). Some less conventional options: [Warning: Some researchy stuff ahead] a) Mazu and Arbor provide products that can detect and optionally shape traffic to avoid DDoS attacks. Must be installed in-line to shape, and can't (AFAIK) shape at really really high line speeds. But for reasonable things like, maybe gigabit and under, I think they can provide pretty reasonable protection. Don't quote me for sure on the rates. b) Ioannidis and Bellovin proposed a mechanism called Pushback for automatically establishing router-based rate limits to staunch packet flows during DoS attacks. [NDSS 2002, Implementing Pushback: Router-Based Defense Against DDoS Attacks] c) I stole some ideas from a sigcomm paper this year (SOS: Secure Overlay Services) to propose a proactive DDoS resistance scheme I term Mayday. The basic idea is that you pick some secret attributes of your packets - destination port, destination address, etc. - and only allow packets with the right values through. You then tell that secret to someone like Akamai, and have them proxy all requests to you. Then you ask your upstream to proactively deny all packets without the magical values. http://nms.lcs.mit.edu/papers/mayday-usits2003.html It's a little weird, but I'd be willing to bet that one of the big overlay providers like Akamai could actually pull it off. The advantage of this approach is that you can implement it without fixing the whole world, unlike egress filters. The downside is that you need someone with lots of nodes. I'd be interested in hearing folk's comments about the mayday paper, btw, since I have to babble about it at a conference in a month. ;-) -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: Is there a line of defense against Distributed Reflective attacks?
On Fri, Jan 17, 2003 at 01:11:14AM -0500, David G. Andersen mooed: b) Ioannidis and Bellovin proposed a mechanism called Pushback for automatically establishing router-based rate limits to staunch packet flows during DoS attacks. [NDSS 2002, Implementing Pushback: Router-Based Defense Against DDoS Attacks] I should have been a bit more accurate here. The proposal for pushback is actually earlier than the implementation paper I cited above: Controlling High Bandwidth Aggregates in the Network. Ratul Mahajan, Steven M. Bellovin, Sally Floyd, John Ioannidis, Vern Paxson, and Scott Shenker. July, 2001. and it also included an internet-draft: http://www.aciri.org/floyd/papers/draft-floyd-pushback-messages-00.txt I believe that Steve Bellovin gave a talk about it at NANOG 21: http://www.research.att.com/~smb/talks/pushback-nanog.pdf -Dave (I'll learn not to send mail past midnight some day) -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: Weird networking issue.
Rule number 1 with any ethernet: Check to make sure you have the duplex and rate statically configured, and configured identically on both ends of the connection. I'd wager you've got half duplex set on one side, and full on the other... -Dave On Tue, Jan 07, 2003 at 02:19:10PM -0500, Drew Weaver mooed: Hi, this is kind of a newbie question but this doesn't make a whole lot of sense :P I have an etherstack hub connected to a FastEthernet port on a cisco 3660 router, these are the stats when I do a show int fast0/0: 5776 input errors, 5776 CRC, 2717 frame, 0 overrun, 0 ignored Whats weird is I just cleared the counters 12 minutes ago, and already there are almost 6000 CRC errors. This connected via a standard Cat5 ethernet cable, I have tried replacing the cable to noavail. Is this a fairly normal situation, If so that's great, but it seemed rather ridiculous to me, and if it is not a normal situation, what would cause this? Any ideas are appreciated. Thanks, -Drew Weaver -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: DNS issues various
On Thu, Oct 24, 2002 at 04:07:18PM -0400, Richard A Steenbergen mooed: We're still working on the distributed attacks, but eventually we'll come up with something just as effective. If it was as easy to scan for networks who don't spoof filter as it is to scan for networks with open broadcasts, I think we'd have had that problem licked too. Are you sure? * A smurf attack hurts the open broadcast network as much (or more) than it does the victim. A DDoS attack from a large number of sites need not be all that harmful to any one traffic source. * 'no ip directed broadcast', which is becoming the default behavior for many routers and end-systems, vs. 'access-list 150 deny ip ... any' 'access-list 150 deny ip ... any' ... 'access-list 150 permit ip any any' (ignoring rpf, which doesn't work for everyone). Until the default behavior of most systems is to block spoofed packets, it's going to remain a problem. -Dave, whose glass is half-empty this week. :) -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: Hunting for bogus BGP announcement for 204.106.93.155
On Thu, Oct 03, 2002 at 06:48:53PM +0200, Jesper Skriver mooed: On Thu, Oct 03, 2002 at 04:35:45PM +0100, [EMAIL PROTECTED] wrote: For the last two days, between approximately 7pm to 2am Eastern time, a spammer hijacked a piece of our address space, presumably by announcing some size of aggregate containing the IP address 204.106.93.155. During the time that the spammer had connectivity using this bogus announcement, RIS didn't pick anything up Nor did our BGP monitors, nor our db of Routeviews. http://bgp.lcs.mit.edu/ Interestingly, we se _no_ announcements of any netblock containing this address, ever. I assume you haven't brought this address space on-line yet? -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: IP over in-ground cable applications.
On Thu, Sep 12, 2002 at 03:04:35PM -0400, Deepak Jain mooed: You would need multicast speakers (routers, etc) along the cable route to effectively multiple your bandwidth at all. Since cable is already multicasting (1 stream to many/all) I don't think I see any advantage. Unless, of course, you expect cable customers to be broadcasting to other cable customers (say their own home video content)... Then MPEG2 Multicast would be your friend. I don't think the answer is as simple as that. It really depends on the number of subscribers per last-hop multicast box, and on the number of channels you offer / popularity distribution of the channels. If you've got 5 channels and 10,000 subscribers per box, multicast saves you nothing. If you've got 1000 channels and 100 subscribers per box, ... -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: Bad bad routing problems?
In the last few days, it's been advertised, but often withdrawn. Perhaps the timing of the announces and withdrawls will help you: MIT saw it advertised via 1 701 8001 4276: http://bgp.lcs.mit.edu/bgpview.cgi?time=dayprefix=216.223.192.0%2F19rel=eqtable=updatesaction=list PSG (Randy Bush's BGP feed) saw it via 2914 8001 4276: http://bgp.lcs.mit.edu/bgpview.cgi?time=dayprefix=216.223.192.0%2F19rel=eqtable=ns0_psg_com_updatesaction=list Playground.net saw it via 14177 15290 7018 701 8001 4276: http://bgp.lcs.mit.edu/bgpview.cgi?time=dayprefix=216.223.192.0%2F19rel=eqtable=nortel_ron_lcs_mit_edu_updatesaction=list ... and so on. (Change the table menu to see a different view) -Dave On Sat, Aug 31, 2002 at 11:25:00AM -0500, Malayter, Christopher mooed: I'm not receiving that network at all. I looked at our border to sprint, genuity, cw, PSI, and our AADS peers. Sorry to not be of more assistance. -Chris -Original Message- From: Mike Tancsa [mailto:[EMAIL PROTECTED]] Sent: Saturday, August 31, 2002 11:14 AM To: Gerald Cc: [EMAIL PROTECTED] Subject: Re: Bad bad routing problems? Strange, from my network I see you via GT and via Telus but not via ATT From me (as11647) BGP routing table entry for 216.223.192.0/19 Paths: (2 available, best #2, table Default-IP-Routing-Table) Not advertised to any peer 852 174 8001 4276 209.115.141.1 from 209.115.141.1 (209.115.137.196) Origin IGP, localpref 100, valid, external Last update: Sat Aug 31 08:47:55 2002 6539 8001 4276 64.7.143.42 from 64.7.143.42 (64.7.143.42) Origin IGP, localpref 100, valid, internal, best Community: 6539:30 Last update: Sat Aug 31 08:47:30 2002 From the ATT CANADA routeserver (route-server.east.attcanada.com) ATT route server route-server.eastshow ip bgp 216.223.192.0/19 % Network not in table route-server.east ATT Canada gets to 8001 via Williams, but I dont see your /19 through them. route-server.eastshow ip bgp regex _8001_ BGP table version is 2266132, local router ID is 216.191.65.118 Status codes: s suppressed, d damped, h history, * valid, best, i - internal Origin codes: i - IGP, e - EGP, ? - incomplete Network Next HopMetric LocPrf Weight Path *i12.129.128.0/24 216.191.64.253 80 0 7911 8001 22420 i *i63.74.146.0/23 216.191.64.253 80 0 7911 8001 23368 i At 10:54 AM 8/31/2002 -0400, Gerald wrote: We are seeing bad routing problems from outside our network. Can anyone corroborate this or help? We are on AS4276 and all traffic from us to our upstream seems good. Great way to spend holiday weekend. /me wonders if anyone is even awake on the NANOG list. :-) 2 addresses in our network I've tested with are: 216.223.200.14 and 216.223.192.68 Many many traces done, some interesting ones below: Here is a succesful traceroute: traceroute to 216.223.200.14 (216.223.200.14), 30 hops max, 40 byte 1 10 ms 10 ms 10 ms rt1.altair7.com [209.11.155.129] 2 10 ms15 ms16 ms 209.10.41.130 3 10 ms16 ms15 ms ge-6-1-0.core1.sjc1.globix.net [209.10.2.193] 4 10 ms16 ms16 ms so-4-2-0.core1.sjc4.globix.net [209.10.11.221] 563 ms46 ms47 ms so-1-0-0.core2.cgx2.globix.net [209.10.10.150] 662 ms63 ms47 ms so-0-0-0.core1.cgx2.globix.net [209.10.10.157] 778 ms94 ms94 ms so-1-0-0.core2.nyc8.globix.net [209.10.10.162] 878 ms94 ms94 ms pos15-0.core2.nyc1.globix.net [209.10.11.169] 978 ms94 ms94 ms s5-0-peer1.nyc3.globix.net [209.10.12.18] 1078 ms94 ms94 ms nyiix.peer.nac.net [198.32.160.20] 1178 ms94 ms93 ms internetchannel.customer.nac.net [209.123.10.34] 1278 ms94 ms94 ms auth-2.inch.com [216.223.200.14] These show failures before reaching our network: traceroute to 216.223.192.68 (216.223.192.68): 1-30 hops, 38 byte packets 1 paleagw4.hpl.external.hp.com (192.6.19.2) 1.95 ms 1.95 ms 5.86 ms 2 palgwb01-vbblpa.americas.hp.net (15.243.170.49) 0.976 ms 0.977 ms 0.976 ms 3 svl-edge-15.inet.qwest.net (65.115.64.25) 0.976 ms !H * 0.977 ms !H 4 * * * 5 * * * traceroute to 216.223.192.68 (216.223.192.68), 30 hops max, 40 byte packets 1 e3-13.foundry1.cs.wisc.edu (198.133.224.116) 2.195 ms 1.754 ms 1.663 ms 2 e15.extreme1.cs.wisc.edu (128.105.1.1) 0.856 ms 0.765 ms 0.737 ms 3 144.92.128.194 (144.92.128.194) 1.550 ms 1.171 ms 1.264 ms 4 * * * 5 * * * All global crossing traces seem to loop within their router: traceroute to 216.223.192.68 (216.223.192.68), 30 hops max, 40 byte 1 64.214.13.1 (64.214.13.1) 0.695 ms 0.889 ms 0.485 ms 0.453 ms 0.350 ms 2 64.214.13.1 (64.214.13.1) 4.535 ms !N ms * ms 0.574 ms !N ms 3 *
Re: Standalone Stratum 1 NTP Server
On Tue, Aug 27, 2002 at 11:07:10PM -0700, Jim Hickstein mooed: --On Wednesday, August 28, 2002 12:51 AM -0400 David G. Andersen [EMAIL PROTECTED] wrote: At work, it's all steel studs and foil-backed wallboard, and the windows (for a patch GPS antenna) are _way over there_. *sigh* I'd love it if someone would pay for my roof penetration there. Does your cell phone work in the room? The CDMA time receivers work in the strangest places. The only places I've had no luck: - In the bowels of a big building along the mass tech corridor - when moved to a bad spot inside a fairly steel and concrete-heavy lab at the university of utah (works in other spots in the lab). But aside from that, I've got them working in network closets and labs all over the place. Worth giving a shot if you're really desperate to play. They're not quite as accurate as GPS (~10 microseconds vs. ~2-5 microseconds), but what's a few microseconds compared to sticking an antenna on the roof? (As a frequency standard, they're quite good. But you can't autocorrect for the CDMA propagation delay). -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: Standalone Stratum 1 NTP Server
On Tue, Aug 27, 2002 at 11:57:39PM -0400, John Todd mooed: Hmm... $2400 is still in the pricey range to be throwing out bunches of these across a network in wide distribution. (Pardon me [...] One would think that a vendor could come up with a 1u rackmount box with a GPS and single-board computer (BSD or Linux-based) for ~$500 total cost. Add 150% for profit and distribution costs, you're still in the $1300 range, which is more reasonable. I suppose my oversimplification is the reason I'm not in the hardware business. You might be imagining a somewhat larger market for standalone stratum-1 timeservers than you might imagine. For real accuracy, you don't want standalone -- you want a locally connected source that you can use to tightly discipline the local clock. (When I say real, I mean sub-milisecond). The difference is .. substantial. Taken from two of my machines on the same subnet: remote st poll reach delay offsetdisp Local CDMA 0 32 377 0.0 -0.11 0.00047 100Mbps Ethernet 1 1024 377 0.00035 0.001103 0.01866 And if you want paranoia, go by ntp's estimate of its accuracy: Local maximum error 5449 us, estimated error 3 us, TAI offset 0 Ether maximum error 584994 us, estimated error 1241 us, TAI offset 0 With that on the board... why do you need, or even want, a standalone NTP server if you're on a budget? Almost certainly you have a local computer in your POP -- you can even get Cisco routers to talk with a local time receiver, if all you want to do is discipline your routers. If you've got a caching nameserver or something else in your POP, that will do just as well. I'd be even happier with a PCI-bus card that I could put into an old (reasonably fast) PC and a CD-ROM with an OpenBSD distribution that automatically did the Right Thing. There is a case to be made about Grab a serial CDMA/GPS unit (I use the EndRun Praecis Ct because it's CDMA; I mention some GPS units below), plug it into your serial port, and stick: server 127.127.29.0 prefer fudge 127.127.29.0 refid CDMA in ntp.conf. It's about as simple as you can get. But remember -- regardless of how nifty your local clock is, you still need to have a good server mesh with NTP. Clocks go bad. CDMA base stations screw up (we've found one so far) or change protocols unexpectedly (three). GPS has serious visibility issues unless you can get an actual roof antenna (two). Configuring this mesh in an intelligent way takes work. Would make a great research project. :) The Ct costs something like $1100. endruntechnologies.com. synergy-gps.com sells a really nice GPS timing unit based on the Motorolla UT+ chipset (designed for timing), including all the parts you need, for .. eh, 600? I forget. Maybe a bit less. Plug into serial port, go. Requires a recompiled kernel under FreeBSD and Linux, but it's fairly easy to set up. If you want something for a bit less work, look at the Trimble units. (For reference: I've got two of the UT+ GPS units, and 20 EndRun Praecis Ct's. Like them both. The Ct is a heck of a lot easier to deploy in a datacenter, as would be the CDMA TrueTime model) If you're really broke, and want a stratum 1 server, host one of our network measurement boxes. We'll ship it to you, you provide the network. In return, you get a local stratum-1 timeserver, managed by yours truly. (I'm serious about this offer, btw.) As a second option: If you manage the connections between your POPs, you can get really decent remote NTP performance. The places in which NTP dies are where latencies are asymmetric. With priority assigned to inter-POP NTP traffic and known symmetric links, life could be quite happy. -Dave (time is very cool) -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/ I do not accept unsolicited commercial email. Do not spam me.
Re: looking glass
On Thu, Jul 18, 2002 at 12:00:38PM -0700, Scott Granados mooed: What are people using for looking glass software. Is it just some simple perl code which grabs data from the router or is it more complex than that? It's just perl. I have a copy of it at http://www.angio.net/rep/lg.tar.gz if you want it. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/
Re: ARIN IP allocation questionn
Technically, you can't sell them to someone else. -Dave On Thu, Jun 27, 2002 at 07:37:34AM -0400, Ralph Doncaster mooed: There's lots of old C's that aren't being announced any more. You might be able to find one that someone can lend you to use. Strangley a search for portable class C on ebay didn't find anything though... Ralph Doncaster principal, IStop.com -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/
Re: Testing Bandwidth performance
On Wed, Jun 26, 2002 at 06:18:00AM -0700, todd glassey mooed: Oh and use something like a SNIFFER to generate the traffic. Most of what we know of as commercial computer's cannot generate more than 70% to 80% capacity on whatever network they are on because of driver overhead and OS latency etc etc etc. It was funny, but I remember testing FDDI on a UnixWARE based platform and watching the driver suck 79% of the system into the floor. Btw, if you've got a bit of time on your hands, the Click router components have some extremely-low-overhead drivers (for specific ethernet cards under Linux). They can generate traffic at pretty impressive rates. They used them for testing DOS traffic for a while. http://pdos.lcs.mit.edu/click/ (Most of the driver overhead you see is interrupt latency; click uses an optimized polling style to really cram things through). Also, the new FreeBSD polling patches should make it so you can get more throughput from your drivers when doing tests. I understand there are similar things for Linux. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/
Re: Global view increase (was:BGP route explosion)
This is more w.r.t. the huge burst of announcements yesterday, not a persistent increase in the routing table sizes, but.. We saw absolutely huge amounts of announcements from 1 3459 17676 (sometimes with padding) For example, see: http://ginseng/bgpview.cgi?time=betweenstart=2002-05-01+06%3A00%3A00end=2002-05-01+07%3A00%3A00bins=100prefix=rel=eqaspath=asrel=containorigin_as=17676scale=lineartable=updates_newaction=plotView=View (Sorry for the long URL). Shows that we received something like 30k announcements that originated from 17676 between 6 and 7 am on May 1st. They were primarily announcing /23s out of large address space ranges allocated to APNIC, like 219.31/16 and friends. -Dave On Thu, May 02, 2002 at 11:33:50AM +0100, Stephen J. Wilcox mooed: We see that too Predominantly seems to be massive amounts of /24s in a couple of nets which were previously /16s. Culprit would appear to be AS705 *i63.0.0.0/24 62.24.196.1 100 0 286 209 701 705 i *i63.1.0.0/24 62.24.196.1 100 0 286 209 701 705 i *i63.2.0.0/24 62.24.196.1 100 0 286 209 701 705 i . Lots of new /20-24 in 67.0.0.0/8 also AS705 In total AS705 is announcing 4432 new routes from yesterday. If you're interested a copy of all new routes since yesterday is at: http://noc.opaltelecom.net/newbgp020502.txt THe file is 0.5Mb so I couldnt really email it :) Steve On Thu, 2 May 2002, James Spenceley wrote: Around 15mins ago an additional ~5,000 routes entered the global view, sadly they appear to be hanging around. Last Tuesday had an increase of 2,000 routes. +7000 routes in a week is significant de-aggregation or leak, any ideas on where these routes are flowing ? I've not seen an increase on any of our peers, so I can only assume its coming from a network who doesn't peer particularly openly. -- James -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/
BGP route update propagation questions
I'm trying to get a better feel for the dynamics of some maybe-necessary BGP routing traffic, and had a few questions: Under what circumstances will BGP send an update (of any sort) to a peer when there is an internal failure that does _not_ result in the complete isolation of a prefix? For example: __ AS1-rtr1 ---*-- AS2-rtr1 --\ / |-- AS3 10/8 -X- AS1-rtr2 ---*-- AS2-rtr2 --/ ^ |--link goes down at X In this case, AS1 is announcing 10/8 to its peer, AS2. An internal link within AS1 goes down. Are there any circumstances under which AS2 will announce some kind of change to AS3? My thoughts: * AS1 might announce a MED change to AS2, and AS2 might propagate that to AS3 in some unknown way (manually?) * AS2 would see the path to AS1 change, and would change its announced MED accordingly * For some reason, AS1 might announce a withdraw on 10/8, and then re-announce it. Is this possible, and under what circumstances would this happen? * Other things? I feel like the answers might depend on networks using ibgp with route reflectors vs. full mesh, or using a different interior routing protocol. In a related situation, let's say AS1 did a 'clear ip bgp' on AS1-rtr2, which was the preferred link for AS2 to get to 10/8. Would AS2 propagate any internal BGP announcements to AS3? (Same kinds of reasons as above, but this time AS1-rtr2 actually stopped announcing anything to AS2) I'm not certain of the right framework in which to think about the answers to these questions, alas. Clues from the operational side would be very welcome indeed. -Dave -- work: [EMAIL PROTECTED] me: [EMAIL PROTECTED] MIT Laboratory for Computer Science http://www.angio.net/