July 1997 - NANOG - lists.nanog.org

PrimeNet's connectivity
by Vincent Poy 27 Jul '97

27 Jul '97

Greetings everyone, Does anyone know how PrimeNet is connected to the net? I know they claim to have a nationwide backbone but they aren't one of the nationwide backbone providers. Cheers, Vince - vince(a)MCESTATE.COM - vince(a)GAIANET.NET ________ __ ____ Unix Networking Operations - FreeBSD-Real Unix for Free / / / / | / |[__ ] GaiaNet Corporation - M & C Estate / / / / | / | __] ] Beverly Hills, California USA 90210 / / / / / |/ / | __] ] HongKong Stars/Gravis UltraSound Mailing Lists Admin /_/_/_/_/|___/|_|[____]

2 2

Re: Non-ISP companies multi-homing?
by Eric Germann 27 Jul '97

27 Jul '97

At 03:27 PM 7/26/97 -0400, Tim Gibson wrote: > > >On Fri, 25 Jul 1997, Eric Germann wrote: > >> Go back and sweep MCI and PSI and Sprint et. al. and SEE what they really >> use. Then follow your own guidelines and maybe the startups won't bitch so >> much... >> >> You can bet when people start paying for addresses there gonna be pretty >> pissed when they aren't fully routable. >> > >Eric, > Ping sweeping will only give NSI/InterNIC a rough estimate of >current online use. A route could be down at the time of sweep on one >end, and on the other a single web box could be using 50 IPs. Changes >would deffinately be welcome in policy. Policing via an "are you awake" >really shouldn't be one of them. > Which is why I agree ping sweeps won't work. And of course, we all keep our SWIP data up to date. BTW, I would hazard a guess that most of the wasted IP space is in the academic community. Does anyone really believe that MIT uses 16 million addresses? And I bet Internic asks for engineering diagrams to review from the big boys whenever they go back for an allocation. Kim, do you folks? I have a university client with 1200 hosts on 4 subnets. Guess what they've got? A /16. % Utilization: 1.83% % Waste: 98.17% Folks thats 64334 addresses. And we try and jam clients in to blocks of 8 or 16 discrete addresses. Squatters rights I guess :) I've seriously thought of approaching them for their "quality" IP real estate which isn't in the swamp. >From a post earlier this week to NANOG: BGP table version is 4043818, main routing table version 4043818 52606 network entries (146045/157874 paths) using <color><param>ffff,0000,0000</param>11039224</color> bytes of memory 12848 BGP path attribute entries using 1589828 bytes of memory 27663 BGP route-map cache entries using 442608 bytes of memory 0 BGP filter-list cache entries using 0 bytes of memory Dampening enabled. 245 history paths, 46 dampened paths Feel free to privately email me ( mailto:ekgermann@cctec.com )as to why if I can jam 128M+ into a backbone router for peering and from above a 53K routing table fits in about 11M, why all the NSP's complain about bloat in the table? Poor lookup algorithm? Input would be appreciated for design references. Mail me and let me know. I don't want to dampen the S/N ratio on NANOG anymore than I have to. >BTW I am one of those that constantly has to scrounge for IP space. > > >Tim Gibson >Skyscape Communications > > ================================================================================ Eric Germann Computer and Communications Technologies ekgermann(a)cctec.com Van Wert, OH 45891 Phone: 419 968 2640 http://www.cctec.com Fax: 419 968 2641 Network Design, Connectivity & System Integration Services A Microsoft Solution Provider

5 6

Re: Non-ISP companies multi-homing?
by Eric Germann 26 Jul '97

26 Jul '97

In honor of Ms. Hubbards pontification, I am starting a little archive: mailto:tcap@cctec.com This archive will be for people to send me (so we don't clog up NANOG) Totally Clueless Allocation Policies of providers. So all you consultants, send me your examples of hosed allocation policies from providers, large and small. You don't have to name your clients, but please do name the providers. I will summarize and forward to Kim and anyone else who cares at the end of two weeks and maybe we'll get an idea of who allocates correctly and who doesn't I'll start off with two examples: 1) Billion dollar company. MCI is their upstream. 5 divisions on one campus. Each has a mailgate, proxy, webserver, and router. Total need per division: 4 addresses. Total allocation: 5 x /24 . % Used: 1.6 % Waste: 98.4 MCI's response when we said take em back: "Why?" 2) Small office. PSI DDR LAN solution. 1 office, 2 PC's, one router. Total need: 3 Total Allocation: 1 x /24. % Used: 1.18% % Waste: 98.82 PSI's response when he said "All I need is a /29": "This is the way we do it. It's just easier for us." BTW Kim, I've gotten 4 responses in two hours on how to nail down all your addresses so a ping sweep shows them in use. Might just be a LITTLE more prevalent than ya'll think, eh? Go back and sweep MCI and PSI and Sprint et. al. and SEE what they really use. Then follow your own guidelines and maybe the startups won't bitch so much... You can bet when people start paying for addresses there gonna be pretty pissed when they aren't fully routable. Anyone care for the summary in two weeks? Mail me privately. Eric At 04:24 PM 7/25/97 -0400, Kim Hubbard wrote: >> >> I'm beginning to think there is a market for a device which has 1 Ethernet >> port and responds to any RANGE of addresses, so you can scam Internic into >> thinking you have 100% utilization of your address space, right off the bat... > >Yeah, because the conservation of IP space only helps InterNIC - it doesn't >help you or your customers or anyone else on the Internet. Let's just >use up all the address space....that'll teach the InterNIC! > >Kim > > > >> >> Eric >> > > > ============================================================================ ==== Eric Germann Computer and Communications Technologies ekgermann(a)cctec.com Van Wert, OH 45891 Phone: 419 968 2640 http://www.cctec.com Fax: 419 968 2641 Network Design, Connectivity & System Integration Services A Microsoft Solution Provider

2 1

Re: Sprint Flapping?
by asmith＠kwiksilver.com 26 Jul '97

26 Jul '97

just a little FYI nobody ever admit's that they are having a failure just that they are working with a new "feature"!:-) At 04:40 PM 7/23/97 -0400, you wrote: >On Tue, 22 Jul 1997, Robert Bowman wrote: > >> According to Sprint NOC at 800-669-8303 they are experiencing routing >> instabilities due to a "Cisco Feature". Not sure what that means. Earlier, > >Hmmm, this occurred pretty soon after Sprint eliminated their 'visible >confederations' hack. >http://www.cctec.com/maillists/nanog/current/msg00113.html > >Maybe they triggered a bug in the route reflector code? > >pbd >-- >Going to church does not make a person religious, nor does going to >school make a person educated, any more than going to a garage makes a >person a car. > > -Al Smith Peace is obtained through superior fire power.

2 1

Ascend and TTL
by Dennis Simpson 26 Jul '97

26 Jul '97

I apologize for not including any business or philosophical perspectives on either DNS or fiber management, but I have just learned something about the little Ascend pipeline boxes that some of our clients use which bothers me. I was wondering if anyone has run across this one before. I checked their web site for tech support, but the goofs don't even provide a search func for technical information. The nanog archive at Merit returns nothing useful for this topic when searched by me. Why do packets going through the pipeline products have their ttl chopped drastically? For example: traceroute to pc1.va.connex.net (205.210.187.50): 1-30 hops, 38 byte packets 1 pro1.barrie.connex.net (205.189.201.1) 1.10 ms 0.795 ms 0.753 ms 2 p25.va.connex.net (205.210.187.49) 28.2 ms (ttl=63!) 27.6 ms (ttl=63!) 67.7 ms (ttl=63!) 3 pc1.va.connex.net (205.210.187.50) 28.7 ms (ttl=30!) 37.6 ms (ttl=30!) 28.9 ms (ttl=30!) The pc on this user's site is a typical win95 box. pro1 happens to be a livingston pm3, but we see this with cisco 1005, 2501 as well. The drop always happens at the Ascend. Another customer has a connection to a different provider and his provider uses an Ascend to connect to my client's ascend, and both routers in the path do this, so his reach onto the net appears to have a very limited distance (hop-wise). Code revs in the pipelines is 4.6C in one case, 4.5patch something in the other. Both exhibit similar behaviour. I don't use Ascend's and have no manuals (learned a long time ago they give me heartburn), so I was hoping someone on this list would be familiar with this behaviour and might be able to suggest how to correct it. Any takers? Thx, dennis

1 0

Kim, where are youuuuuuuuuuu!
by Eric Germann 26 Jul '97

26 Jul '97

Fitting... Maybe they yanked their own blocks. Kim, send me an address where I can forward the study to. Sorry for the noise. >From: kimh(a)internic.net >Date: Fri, 25 Jul 97 19:16:56 PDT >To: ekgermann(a)cctec.com >Subject: Your mail to Kim Hubbard <kimh(a)internic.net> > >Your message was delivered to an email destination which is >no longer in service. > >Please contact Energetic Systems <info(a)energetic.com> for more information. > ============================================================================ ==== Eric Germann Computer and Communications Technologies ekgermann(a)cctec.com Van Wert, OH 45891 Phone: 419 968 2640 http://www.cctec.com Fax: 419 968 2641 Network Design, Connectivity & System Integration Services A Microsoft Solution Provider

2 1

RE: nanog mailing list
by Steve Carter 26 Jul '97

26 Jul '97

I have no real idea why this is, but this industry seems to breed more than it's fair share of arrogant idiots. I saw, at least, one of the offensive responses to your request and I assume, having a naive inflated opinion of human nature, that this person has an esoteric sense of humour and wasn't meaning to be arrogant or offensive at all. Even this person, who obviously knows more about everything than everyone else, had to start somewhere and I hope he got a better welcome than you. > My apologies to the wide variety of people aparently offended by my > request for subscription, and many thanks to the people who pointed me > > towards the proper subscription channels. > > I recall being told this list was private and required references, my > mistake. > > Jason Storm

2 1

7k extra prefixes?
by Alec H. Peterson 25 Jul '97

25 Jul '97

Can someone explain this: BGP table version is 4043818, main routing table version 4043818 52606 network entries (146045/157874 paths) using 11039224 bytes of memory 12848 BGP path attribute entries using 1589828 bytes of memory 27663 BGP route-map cache entries using 442608 bytes of memory 0 BGP filter-list cache entries using 0 bytes of memory Dampening enabled. 245 history paths, 46 dampened paths Note the second line. That's up about 7k prefixes from what it was last time I looked. I'm looking through the routing table now and haven't found anything obvious yet. Anybody? Alec -- +------------------------------------+--------------------------------------+ |Alec Peterson - ahp(a)hilander.com | Erols Internet Services, INC. | |Network Engineer | Springfield, VA. | +------------------------------------+--------------------------------------+

1 1

Re: Fiber Cut In Sterling VA.
by Steve Miller 25 Jul '97

25 Jul '97

Hey... Get back to work! How's life in ISP land? Hope all is well. Steve At 10:18 AM 7/25/97 -0400, Henry Clark wrote: >At 04:18 AM 8/25/97 -0400, Shaun S. Amini wrote: >>And the trend continues... We had 2 major fiber cuts here @ 1:00 P.M. >>thanks to Joy Construction who was digging for MCI Metro. > >duck season >rabbit season >duck season >rabbit season >duck season >rabbit season > ><evil look> > >backhoe season :-) > >

1 0

RE: Non-ISP companies multi-homing?
by Eric Germann 25 Jul '97

25 Jul '97

At 11:22 AM 7/24/97 -0400, root(a)gannett.com wrote: >On Thu, 24 Jul 1997, Howard C. Berkowitz wrote: > >> Interesting approach. In general, the ISPs I know would be reluctant to >> run iBGP with a customer, unless they had total control of all BGP >> speakers. If I understand you correctly, the enterprise would have to tag >> its advertisements to the second ISP with the ASN of the first, since the >> enterprise doesn't have its own. Again, I think most ISPs would be >> reluctant to give up this amount of control. > >I think most of the companies running redundant links now have their own >address space and ASN. We got our primary address blocks back when a >company could still do that. I think there's going to have to be some >way to address that with semi-portable AS' in the near future though, as >more criticality transitions to the Net. > So how do you folks punch through the infamous /19 filters? I've got a couple of clients who would like to multihome, but can't get PI space cause they can't justify 8192 addresses. Not having "fully routable" PI space negates the whole purpose of multihoming from their perspective. Does Gannett or Pointcast have >= 8K hosts exposed on their DMZ networks? I'm beginning to think there is a market for a device which has 1 Ethernet port and responds to any RANGE of addresses, so you can scam Internic into thinking you have 100% utilization of your address space, right off the bat... Eric

6 5