mailing list archives
Re: Non-ISP companies multi-homing?
From: Eric Germann <ekgermann () cctec com>
Date: Sat, 26 Jul 1997 16:23:55 -0400
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
>> You can bet when people start paying for addresses there gonna be pretty
>> pissed when they aren't fully routable.
> 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 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
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.
Eric Germann Computer and Communications Technologies
ekgermann () 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
- RE: Non-ISP companies multi-homing?, (continued)