From: Howard C. Berkowitz (hcb@gettcomm.com)
Date: Tue Mar 11 2003 - 15:40:27 GMT-3
At 10:09 AM -0500 3/11/03, Peter van Oene wrote:
>Hi all,
>
>Here is a quick post from Dave Katz on ISIS vs OSPF in large
>networks dealing with the issue of which protocol inherently scales
>better. This is from a thread in the IETF OSPF WG mailing list for
>those looking for the full thread. Dave has participated
>significantly in the development of routing protocol software for
>both Cisco and Juniper.
>
>Thought some folks might find it interesting
>
>Pete
As far as the "implementations of ISIS from various vendors," Dave
wrote all of the ISIS code involved. There was turnover in Cisco's
early OSPF developers.
>
>
>>Date: Sun, 9 Mar 2003 21:05:14 -0800
>>Reply-To: Mailing List <OSPF@DISCUSS.MICROSOFT.COM>
>>Sender: Mailing List <OSPF@DISCUSS.MICROSOFT.COM>
>>From: Dave Katz <dkatz@JUNIPER.NET>
>>Subject: Re: ospf limits...
>>To: OSPF@DISCUSS.MICROSOFT.COM
>>X-RAVMilter-Version: 8.4.1(snapshot 20020919) (usermail.com)
>>X-Spam-Status: No, hits=-8.5 required=9.1
>> tests=FORGED_RCVD_TRAIL,IN_REP_TO,REFERENCES
>> version=2.50
>>X-Spam-Level:
>>X-Spam-Checker-Version: SpamAssassin 2.50 (1.173-2003-02-20-exp)
>>
>>For all practical purposes, the designs of the OSPF and ISIS protocols
>>will not be the limiting factor in the size of an area, unless (a) you
>>have a really good implementation, and (b) you feel the need to dump
>>excessive numbers (many thousands) of external and stub routes into
>>the protocol.
>>
>>Most implementations will crash and burn before the topology gets
>>big enough to become an issue, and most people don't dump externals
>>into their IGPs (they use BGP instead.)
>>
>>Architecturally, OSPF limits the inter-router topology and stub routes
>>due to the 64KB limit on the Router LSA, and ISIS limits the total amount
>>of information due to the 256 LSP "fragment" limit. One could come up
>>with various hacks for either protocol if these limits were actually,
>>well, limiting, but this has never been the case in (sane) practice.
>>
>>Historically, the ISIS implementation from a particular major vendor has
>>had better scaling characteristics than the OSPF implementation of that
>>particular major vendor, but this this isn't really the case for another
>>major vendor. ;-)
>>
>>--Dave
This archive was generated by hypermail 2.1.4 : Sat Apr 05 2003 - 08:51:37 GMT-3