[36955] in North American Network Operators' Group
Re: RIP and RIPv2, "The glue that makes the internet work"
daemon@ATHENA.MIT.EDU (bmanning@vacation.karoshi.com)
Thu Apr 26 19:52:37 2001
From: bmanning@vacation.karoshi.com
Message-Id: <200104270011.AAA04291@vacation.karoshi.com>
To: fenner@research.att.com (Bill Fenner)
Date: Fri, 27 Apr 2001 00:11:02 +0000 (UCT)
Cc: nanog@merit.edu
In-Reply-To: <200104262322.QAA22781@windsor.research.att.com> from "Bill Fenner" at Apr 26, 2001 04:22:58 PM
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Errors-To: owner-nanog-outgoing@merit.edu
>
>
> > What was/is the largest production network (in number of end nodes) that
> > used/uses RIP as the IGP?
>
> Xerox routed a few thousand subnets of 13/8 with RIP (v1!) as late as 1998.
> Dunno if that's large enough.
>
> Bill
>
Its pretty big. Most of the data is not verifiable, but
a profile of 100-10,000 subnets, between 5-2000 nodes per subnet
w/ RIPv1 seems to be emerging for sites like Xerox as well
as old NSFnet regionals.
On the other hand, reports of large, multinational networks
running static routing in their cores seem to indicate a
desire to have routing in the core more stable than any dynamic
protocol will allow.
With the growth in the number of injected prefixes and varient
paths, one might say that the "value proposition" of dynamic
routing is not what it once was... Or it could be that the
folks running the big networks are more comfortable with
manual/static routing systems? Such environments certainly
provide easier means to set enforcable service level agreements.
But my muse has gotten the better of me.
--bill
