[85453] in North American Network Operators' Group
Re: Operational impact of depeering
daemon@ATHENA.MIT.EDU (Tom Vest)
Mon Oct 10 14:09:49 2005
In-Reply-To: <A206819EF47CBE4F84B5CB4A303CEB7A01444BA1@dul1wnexmb01.vcorp.ad.vrsn.com>
Cc: Martin Hannigan <hannigan@verisign.com>
From: Tom Vest <tvest@pch.net>
Date: Mon, 10 Oct 2005 14:09:16 -0400
To: Nanog Mailing list <nanog@merit.edu>
Errors-To: owner-nanog@merit.edu
On Oct 10, 2005, at 11:02 AM, Hannigan, Martin wrote:
> Anti-Level(3)? The only fact in this was the route view
> count, and even that could be wrong. Not a very fair
> comparison, especially to make to regulatory people who
> may not know better.
How in the world does this read as anti-Level3?
What precisely is unfair about the comparison?
Concrete suggestions about how to make a fairer comparison,
independently, using public domain information, would be welcome.
> AS 174 was old when it was PSI. It's now Cogents ASN via acquisition.
> You fairly imply that Cogent is as old as PSI in garnering sympathy
> for
> them being old school. Cogent is not old school.
The implication I was making (maybe too subtly) was that counting
this way involves some obvious error terms, one of which is age of
allocation -- meaning specifically, address allocation policies in
effect at the time that the relevant netblocks were allocated. Scale,
a.k.a. host density might be another obvious one. What I meant to
suggest was that this method might overstate Cogent's significance.
However I was wrong. I was thinking of the vintage 1991 PSI "Class A"
that Cogent still routes. I should have said "both old networks,"
given L3's two even older BBN "Class As." But if age tends to suggest
a certain (freedom of) slack in utilization, then that would mean
that the count actually overstates L3's operational significance. So,
does that correction makes it less biased -- or more Anti-Level3?
TV
