May 2002 - Test - lists.nanog.org

Re: Routers vs. PC's for routing - was list problems?
by Vinny Abello 23 May '02

23 May '02

At 04:17 PM 5/23/2002 -0400, you wrote: > > I agree with you on that. Hot swapability for various interfaces is > > something routers obviously have over PC's. > >Hot swap PCI is old news. True, but not widely implemented in the standard PC market. If you want a server that has hot swap capability, you're likely paying a premium price for a lot of extra other features. It's not something you can typically just build yourself, and if you can you'll need a case that allows you easy access to swap the PCI cards. By the time you pay for an enterprise level server with this capability, I would rather have put the money towards a good router. > > True... unless going for 64 bit PCI at 66MHz... still it's obvious that > > routers are designed for one simple purpose and generally have larger > > backplanes to handle that. > >However, $ for $, even when buying used cisco gear at 80% off from >dot-booms, a PC router will outperform any traditional router. At what speeds though? As you get into the higher gbic speeds, a PC doesn't have the backplane to cut it. Now if we're talking raw processing power, a PC can blow away a router in calculations per second any day. :) > > I agree a router is probably more efficient in just routing packets, but in > > complex filtering or traffic manipulation/packet sniffing, a PC might have > > the edge. :) > >Yes, ipfw/dummy is very very cool. Like, inducing a few 100 msecs of >latency to folks who don't pay on time :) Hehehehe... Interesting approach. I find it more fun to just shut them off. It makes them take you more seriously. Unfortunately I would say only a small percentage of users, may 20% or so would even notice the latency issues if they were having them. They're more likely to complain about slow transfer speeds. That is even more fun and can be done on any traditional Cisco... Traffic shaping is cool but hindered by being limited to controlling outbound traffic on an interface. Rate limiting even more fun. Hmm... [exceed action drop] Why is there so much damn packet loss on my connection when I put traffic across it??? ;) Vinny Abello Network Engineer Server Management vinny(a)tellurian.com (973)300-9211 x 125 (973)940-6125 (Direct) Tellurian Networks - The Ultimate Internet Connection http://www.tellurian.com (888)TELLURIAN

1 0

RE: operational: icmp echo out of control?
by James Smith 23 May '02

23 May '02

> -----Original Message----- > From: Richard A Steenbergen [mailto:ras@e-gerbil.net] > Sent: Thursday, May 23, 2002 4:36 PM > To: Mark Kent > Cc: nanog(a)merit.edu > Subject: Re: operational: icmp echo out of control? > > Path latency doesn't change much, you can determine this with very few > probes. Reachability does not need to be continuously probed, > you can take > cues from other data to decide if you need to re-probe. I wonder if this can be used to profile a network for the sales droids? If one of your routers is being pinged continuously, this might be an indication they are using some sort of "Route Optimization/Failover" box. Especially if you ask them why they are pinging you on the order of 1-5pps and they can't really give an answer and don't turn it down. Sic-em! Sell them that managed BGP! For instance, the FatPipe box does this sort of thing "so you don't have to use BGP" to (sort of) multihome. Link failure is detected by loss of ping response. Failover (link and DNS) is under five seconds, so to prevent premature failover, ping often, and only failover if you take, say, three lost packets. The question for network operators is (just so this is kind of OT), is this kind of monitoring by the customer more of an annoyance, or a real item of operational concern? How would you react if you had a customer pinging the router on your side of his link (or farther upstream) that refused to limit the pings? James H. Smith II NNCDS NNCSE Professional Services - Network Engineer The Presidio Corporation

1 0

Re: Routers vs. PC's for routing - was list problems?
by E.B. Dreger 23 May '02

23 May '02

JC> Date: Thu, 23 May 2002 15:25:14 -0400 (EDT) JC> From: James Cornman JC> We've had some rather good success with FreeBSD based PC JC> Routers. Typical setup was FreeBSD 4.x, 512mb, 20gb RAID-1, JC> 3com Gigabit Ethernet card, Fore Systems OC3 ATM card. All JC> this, with zebra on top. It worked well for a long time, JC> although it turned out getting deprecated because of some JC> zebra issues (with ospfd. They (the problems) weren't JC> confirmed by the zebra community but thats the only thing we JC> could narrow it down to. ospfd would die periodically.) The JC> line cards were bought off of eBay. Yes, for <= 155 Mbps, it works well. My intended point was that Juniper != PC. Yes, both are FreeBSD on x86, which works great. But PCs use the system bus, which is a much harsher limit than having a fast backplane or midplane that just switches data. As Randy said, a router must route _and_ forward. When PCI runs out of gas, you just can't push any more through it. Again: Anyone played with cPSB yet? It looks very promising... The "sweet spot" for building a PC-based router probably would be around 2x or 3x DS3 right now. 7200s have come down in price, but DS3 cards are still fairly valuable. (Not enough price difference in the DS1 game to make a PC-based router worth the effort on the low end... unless one is multihoming and needs more RAM than 26xx or 36{20|40} can hold.) I'm trying to remember what "Buy It Now" was on that M20 on eBay the other day... IIRC, it had 4x OC3 + 4x DS3 + 4x FE. JC> We did VLAN trunking through the 3com GBE card to a Catalyst JC> 3548. Did any rate limiting with DUMMYNET and ipfw pipes. JC> Overall, the whole system worked great for a few months JC> without human interaction, until the ospfd problems. How long ago was this? Zebra has been stagnant for nearly a year now, and my recollection was that late 2000 was when OSPF bugs were biting... -- Eddy Brotsman & Dreger, Inc. - EverQuick Internet Division Phone: +1 (316) 794-8922 Wichita/(Inter)national Phone: +1 (785) 865-5885 Lawrence ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Date: Mon, 21 May 2001 11:23:58 +0000 (GMT) From: A Trap <blacklist(a)brics.com> To: blacklist(a)brics.com Subject: Please ignore this portion of my mail signature. These last few lines are a trap for address-harvesting spambots. Do NOT send mail to <blacklist(a)brics.com>, or you are likely to be blocked.

2 1

Discussion of Results
by John Palmer (NANOG Acct) 23 May '02

23 May '02

Proposal #1 (which passed by over 2/3rds - 67.9%) expresses the sense of the GA that DOC should re-bid the ICANN contract and forget ICANN completely Proposal #2 (which passed by 75%) expresses to ICANN the desire that they reform in a meaningful way, and if they don't, that the DOC should replace ICANN. Interesting.... ---- AGN Domain Name Services, Inc http://www.adns.net Since 1995. The Registry for .AMERICA, .EARTH, .LION, .USA and .Z Define yourself or Be Defined. Censorship-free GA list at : http://dns-o.org/mailman/listinfo/ga

2 1

Re: Routers vs. PC's for routing - was list problems?
by James Cornman 23 May '02

23 May '02

We've had some rather good success with PC based routers. Typical setup was FreeBSD 4.x, 512mb, 20gb RAID-1, 3com Gigabit Ethernet card, Fore Systems OC3 ATM card. All this, with zebra on top. It worked well for a long time, although it turned out getting deprecated because of some zebra issues (with ospfd. They (the problems) weren't confirmed by the zebra community but thats the only thing we could narrow it down to. ospfd would die periodically.) The line cards were bought off of eBay. We did VLAN trunking through the 3com GBE card to a Catalyst 3548. Did any rate limiting with DUMMYNET and ipfw pipes. Overall, the whole system worked great for a few months without human interaction, until the ospfd problems. Feel free to contact me off list if you have any questions. I dont know all of the exact hardware/software tweaking that were done; alot of them were left default, but i'll try to help. -- James Cornman <jamesc(a)nac.net> Net Access Corporation - http://www.nac.net/ On Thu, 23 May 2002, E.B. Dreger wrote: > > ADC> Date: Thu, 23 May 2002 14:30:16 -0400 > ADC> From: Anthony D Cennami > > > ADC> "Not to say you can't route well with a linux or bsd system > ADC> you can but at the high-end probably not as well." > ADC> > ADC> Tell that to Juniper. > > Where can I buy their line cards for my PC? > > > -- > Eddy > > Brotsman & Dreger, Inc. - EverQuick Internet Division > Phone: +1 (316) 794-8922 Wichita/(Inter)national > Phone: +1 (785) 865-5885 Lawrence > > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > Date: Mon, 21 May 2001 11:23:58 +0000 (GMT) > From: A Trap <blacklist(a)brics.com> > To: blacklist(a)brics.com > Subject: Please ignore this portion of my mail signature. > > These last few lines are a trap for address-harvesting spambots. > Do NOT send mail to <blacklist(a)brics.com>, or you are likely to > be blocked. > >

1 0

RE: Certification or College degrees? Was: RE: list problems?
by Rowland, Alan D 23 May '02

23 May '02

While the effectiveness of degree requirements may be argued, they are efficient. When your HR department gets hundreds or thousands of applications, they need some way to find the wheat. The net sector is young and was mostly immune to traditional business practices. Not all traditional business practices are bad (see dot.bomb). Lack of business acumen means the days of six figure income and significant stock options because there were 10 job openings for every geek who could RTFM are over. Even though the "job market is coming back" there's still 20 'techies' in Birkenstocks and Star Wars t-shirts for every (decent) job hiring. Everything else being equal (which is often the case) a cert or degree is a great tie-breaker. Welcome to the traditional job market fellow geeks. Remember all the jokes about Sanitation Engineers? ;) Put another way, when you take that expensive car of yours in for service (you do have one if you're successful in this industry, right? ;) ), do you go to Joe's Garage (apologies to all named Joe) or a dealer/service center with "certified" mechanics? Just my 2¢. The delete key is your friend. Best regards, _________________________________________ Alan Rowland (BS in Business and Management, UofM, 1990 "no warranty expressed or implied, use at your own risk, may be terminated at any time without notice" -----Original Message----- From: Christopher J. Wolff [mailto:chris@bblabs.com] Sent: Wednesday, May 22, 2002 11:16 AM To: nanog(a)merit.edu Subject: Certification or College degrees? Was: RE: list problems? I would add to that statement: Requiring a technology certification is equally as obsurd. I've been told I could pass the Emperor-Level CCIE test; however, I do not believe it will add more value for my customers. Regards, Christopher J. Wolff, VP CIO Broadband Laboratories http://www.bblabs.com Andrew Dorsett said: *jumping on my soap box* I have to say that the idea of requiring a degree for the IT industry is obsurd.

14 14

RESULTS: Vote on Two motions about ICANN Reform May 2002
by John Palmer (NANOG Acct) 23 May '02

23 May '02

DNSO General Assembly RESULTS of vote on Two motions about ICANN Reform, May 2002 23 May 2002 RESULTS of vote on Two motions about ICANN Reform, May 2002 Re: b12 http://www.dnso.org/dnso/notes/2002.GA-ICANN-Reform-motions.html Fundamentals - Two Rules for votes by the GA: A. Voting Rules adopted by the GA on 6 July 2000: http://www.dnso.org/dnso/notes/2000.GA-voting-rules.html B. Election Rules (Instant Runoff) adopted by the GA on 23 March 2001: http://www.dnso.org/dnso/notes/2001.GA-chair-election-rules-v0.2.html Vote Tally: 1. Electorate: 586 persons in Voting Roster, when the ballot was prepared http://www.dnso.org/secretariat/b12.rosterindex.html 2. Number of ballots received: 218 , Number of casted votes: 433 3. Participation: 37.2% 4. Vote results: Numbers: 148 I FOR Motion 1 ("Request to US DoC") 54 I vote AGAINST Motion 1 15 I ABSTAIN regarding Motion 1 164 I vote FOR Motion 2 ("Reform principles") 33 I vote AGAINST Motion 2 19 I ABSTAIN regarding Motion 2 --------------------------------------------- 433 TOTAL casted votes Percentages: 67.9% : I FOR Motion 1 ("Request to US DoC") 24.8% : I vote AGAINST Motion 1 6.9% : I ABSTAIN regarding Motion 1 75.2% : I vote FOR Motion 2 ("Reform principles") 15.1% : I vote AGAINST Motion 2 8.7% : I ABSTAIN regarding Motion 2 5. Vote records: The calculations using Voting Rules are published in: http://www.dnso.org/secretariat/b12.fullrecord.html.

1 0

Who posts to the nanog list -- The top 59 players (Was not: Re: list problems?)
by Martin J. Levy 23 May '02

23 May '02

Hi all, Somewhat related to the previous subject, but bringing it much closer to the operational aspects of the nanog e-mail list, I present this data.... It only took a few commands to build a count of articles posted to the nanog list over the last 17 months (Jan 2001 to the present day). I sorted them based upon the total count of messages (highest first). I only included the first 59 names. If the e-mail shows up with wrapped lines, then you can find the same information at http://www.mahtin.com/nanog-count.txt which should be easy to read in an unwrapped manner. Note that as some people posting using different real-names in the SMTP headers, there are minor errors in this list. ( "bmanning" missed counting some of "Bill Manning" or "bill manning" posted messages and "Richard A. Steenbergen" and "Richard A Steenbergen" are really the same person). I can't help it if people change their e-mail setups. ;-) Enjoy the data. Martin ----------------------------------------------------- # E-MAIL NAME COUNT 2001-01 2001-02 2001-03 2001-04 2001-05 2001-06 2001-07 2001-08 2001-09 2001-10 2001-11 2001-12 2002-01 2002-02 2002-03 2002-04 2002-05 1 Sean Donelan 534 68 34 17 44 16 51 24 20 91 36 38 27 6 13 24 17 8 2 Roeland Meyer 371 44 13 51 37 64 - 38 63 61 - - - - - - - - 3 Valdis.Kletnieks 231 14 9 20 6 23 12 15 29 18 22 8 7 7 10 6 10 15 4 Randy Bush 214 18 14 10 6 10 21 6 28 33 13 14 11 9 6 4 5 6 5 John Fraizer 205 18 32 19 31 26 8 10 24 37 - - - - - - - - 6 E.B. Dreger 147 - - - - 12 15 1 8 19 22 5 5 11 1 9 12 27 7 Paul Vixie 128 13 8 1 - 9 7 1 4 12 22 9 - 1 5 4 16 16 8 Shawn McMahon 127 15 16 34 14 48 - - - - - - - - - - - - 9 Richard A. Steenbergen 127 17 8 5 26 23 40 8 - - - - - - - - - - 10 Dan Hollis 127 6 2 4 14 29 4 8 6 10 5 3 1 8 4 1 1 21 11 David Schwartz 122 - 2 32 13 22 12 4 10 8 3 1 - 1 4 3 3 4 12 Greg A. Woods 115 20 14 16 21 8 18 2 5 1 3 - - - - - - 7 13 Scott Francis 113 - 1 25 4 16 2 8 3 4 3 2 6 2 4 1 3 29 14 bmanning 108 15 5 3 9 14 2 4 6 14 4 7 - 6 3 4 8 4 15 Sean M. Doran 106 - - - 9 15 7 - 24 25 6 3 - 1 2 8 5 1 16 Stephen J. Wilcox 105 3 6 10 3 11 3 10 3 10 - - - 1 10 9 6 20 17 Daniel Golding 101 1 5 3 1 7 8 11 6 15 7 9 5 3 4 3 9 4 18 Leo Bicknell 100 3 3 - 1 - 9 5 28 23 12 4 2 2 1 2 1 4 19 Hank Nussbacher 100 12 14 10 7 2 6 5 4 14 8 4 3 1 - 3 2 5 20 Patrick Greenwell 98 7 7 30 2 2 1 1 17 10 4 8 7 2 - - - - 21 Alex Rubenstein 98 1 7 2 2 9 16 7 3 18 4 8 - 3 5 2 3 8 22 Vadim Antonov 94 3 1 12 - 2 - - 25 22 6 3 1 - 13 3 1 2 23 Bill Woodcock 93 9 8 3 5 11 5 10 1 4 3 2 1 5 2 8 10 6 24 Steven J. Sobol 92 4 6 3 1 1 8 8 16 14 8 - 1 4 - 4 3 11 25 Richard A Steenbergen 92 - - - - - - - - - - - - - 2 19 27 44 26 Christopher A. Woodfield 92 2 - 3 1 11 20 12 13 - 12 5 3 3 7 - - - 27 mike harrison 89 - 3 5 16 12 9 5 10 10 5 2 4 - 1 3 2 2 28 Mitch Halmu 88 - - - - 30 3 14 23 2 - - - - - - 1 15 29 Marshall Eubanks 88 10 2 9 5 3 3 11 5 23 4 2 - - - - 6 5 30 Jared Mauch 88 7 5 4 2 5 7 2 2 8 7 5 3 9 10 3 4 5 31 Deepak Jain 87 3 7 2 1 2 14 4 2 14 2 4 - 3 10 9 7 3 32 Steve Sobol 85 12 13 10 6 34 3 - 1 - - - - - 1 4 1 - 33 David Lesher 85 9 10 5 2 11 6 3 4 13 4 1 3 1 4 2 7 - 34 Steven M. Bellovin 84 10 6 14 - 1 1 3 9 13 6 5 2 4 5 4 1 - 35 Jim Mercer 80 3 2 6 13 12 11 3 6 11 - - - - 6 1 5 1 36 Alex Bligh 76 3 3 - 1 6 3 3 18 21 7 10 1 - - - - - 37 Eric A. Hall 73 10 6 9 6 14 4 - 3 6 1 - - 5 3 - 1 5 38 jlewis 69 9 5 4 1 5 1 3 16 4 3 2 - 1 5 2 4 4 39 Stephen Griffin 69 1 6 1 6 5 4 3 - - 1 - 1 10 7 1 3 20 40 Iljitsch van Beijnum 69 - - - - - - - 8 7 21 7 - - 5 5 7 9 41 Patrick W. Gilmore 65 1 - - - 3 2 - 2 40 13 2 - - - 2 - - 42 measl 64 - - - - - - - 9 16 4 9 4 1 4 3 7 7 43 Mike Batchelor 63 - - 17 3 4 1 5 7 5 16 2 2 - 1 - - - 44 Adam Rothschild 63 7 11 5 1 10 4 3 6 4 4 - - 1 1 1 2 3 45 Charles Sprickman 58 3 3 6 2 5 5 5 8 13 3 3 2 - - - - - 46 Joe Abley 57 - 2 4 3 2 14 - 1 5 5 1 1 8 2 4 3 2 47 Matt Levine 56 7 1 4 5 2 9 3 2 9 4 4 1 2 - 1 2 - 48 Majdi S. Abbas 56 2 5 2 7 4 5 9 7 7 - 2 2 1 - 1 1 1 49 Christian Kuhtz 56 1 1 1 2 2 5 5 18 11 3 3 1 - - 1 2 - 50 Simon Lockhart 55 8 7 4 1 3 7 2 - 7 4 - 1 2 - - 4 5 51 Adrian Chadd 55 13 10 7 6 8 - - - - - 3 - 1 - - 7 - 52 Greg Maxwell 54 - - - 19 10 - 3 14 1 2 - - - 2 2 1 - 53 William Allen Simpson 53 2 2 7 - 6 3 10 2 - 8 1 - - 3 9 - - 54 John Payne 52 10 6 11 - 10 2 1 1 4 5 1 - - - 1 - - 55 alex 51 - 1 3 9 2 2 10 2 10 5 - 4 - - 2 - 1 56 Adam McKenna 50 5 4 4 1 6 1 7 1 5 7 6 - - - 1 - 2 57 Simon Higgs 49 - - 19 - - - - - - - 6 4 6 6 2 4 2 58 Mikael Abrahamsson 46 6 2 - 2 6 11 2 5 4 4 2 - - - 1 - 1 59 Ralph Doncaster 45 - - - - - - - - - - - - - - - - 45

3 2

Strange Bandwidth drop: 5/21 14:00 to 5/22 02:00: Any one else see it
by Larry Rosenman 23 May '02

23 May '02

Looking at our graphs, we saw a very significant drop in our inbound bandwidth from Tuesday, 21/May/2002 14:00 (UTC -0500) to 22/May/2002 02:00 (UTC -0500). We can't explain it from internal sources. Did anyone else see this? Does anyone have an explanation? Thanks, LER -- Larry Rosenman http://www.lerctr.org/~ler Phone: +1 972-414-9812 E-Mail: ler(a)lerctr.org US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749

2 1

Re: Certification or College degrees? Was: RE: list problems?
by Petr M. Swedock 23 May '02

23 May '02

: : Vadim Antonov <avg(a)exigengroup.com> : : : > On Wed, 22 May 2002, Kristian P. Jackson wrote: : > : >> Perhaps a bachelors in network : >> engineering is in order? : : I'm afraid there's not enough stuff one has to know to sucessfully : "design" networks to fill more than one-semester course. One should not design networks. Designing networks is not a solitary activity. One should be part of a team that works to design networks. That team should be made up of people who know many and divers things. You are not far off the mark, however: There's not enough stuff, not contained in other coursework, to fill more than one-semester course in networking. That is just to say that 'networking' is not a fundamental discipline in the same way as math, or chemistry might be but is an amalgam of intersecting skills and disciplines that are easily found elsewhere. The engineers I admire and respect all combine analytical skills (math), understanding of the physical layer (electrical engineering, physics), financial and resource allocation (economics), building fitout (mechanical engineering, architecture) and computers (those things that sit at the nodes of the networks...) AND they are constantly striving to better understand the relationships between all those things. None of the engineers I respect and admire, came by the depth and breadth of their skill not by walking a linear path but rather by wandering hither and yon... cataloguing as they went. School made sense for many, less so for others... The pivotal advice I received was regarding a decision to stay, or leave, the small college I was attending and to get an engineering degree at a larger college. My physics professor argued thusly: "You can get and engineering degree and be an engineer, or you can get a physics degree and be anything." Peace, Petr -- petr(a)mit.edu http://web.mit.edu/petr http://lids.mit.edu ____________________________________________________________ You can design simply, or you can design for simplicity. The first requires a fear of complexity only. The second requires an understanding of complexity. Choice is yours ------------------------------------------------------------

1 0