From: Howard C. Berkowitz (hcb@gettcomm.com)
Date: Tue May 11 2004 - 17:14:44 GMT-3
At 3:25 PM -0400 5/11/04, Volkov, Dmitry wrote:
> > -----Original Message-----
>> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
>> Howard C. Berkowitz
>> Sent: May 11, 2004 3:15 PM
>> To: ccielab@groupstudy.com
>> Subject: Re: OSPF route selection
>>
>> The OSPF working group official page is
>> http://www.ietf.org/html.charters/ospf-charter.html
>>
>> This will give you the current standard:
>> http://www.ietf.org/rfc/rfc2328.txt Other documents will give you
>> some special cases.
>>
>> The basic rule:
>> Always prefer intra-area over inter-area over external 1
>> over external 2.
>> Within a category, take the longest match.
>> To break ties with equal match length, use route cost.
>> If there are still ties, load-share up to the configured limit of
>> parallel routes.
>
>I'm wondering If there are let say 5 of equal cost routes within
>category.
>Which 4 (by default) will be taken into RT ?
I'm trusting to wetware here, but I believe, for IGPs, it's generally
the first four recognized. Alex Zinin's book might have the specific,
or I could check with some internal people.
This may be release specific. In contrast, read up on BGP
deterministic metric, which puts more specific controls on what is
taken in what order. I don't think the current implementation of IGPs
can be set to most-recent-preferred.
There aren't strong reasons to take the most recent, other than
perhaps to make debugging a little easier. Remember, these are
already equal-cost routes, so there shouldn't be any benefit to one
over another. If one became less preferable, it would be withdrawn,
and then, indeed, would be replaced by the next new equal cost route.
This archive was generated by hypermail 2.1.4 : Wed Jun 02 2004 - 11:12:10 GMT-3