From: Nigel Roy (nigel@xxxxxxxxxxxxxxxxx)
Date: Wed May 29 2002 - 07:53:55 GMT-3
Guys,
I feel I should also point out here that Cisco have always advised strongly
against using anything other than the defaults of bandwidth and delay for
calcualting metrics. The reasoning behind this is that if you bring load
into the equation the tendancy is to flip flop between the two different
routes. i.e. when the load goes up on one link the metric goes up therefore
more traffic or all traffic(depending on whether variance is used) is
diverted down the other link, then that links load value goes up so it swaps
back.
I did attempt to use this for a customer going back about 6 or 7 years now
and found that it actually did not work on that particular version of
software i.e. the metrics once calculated never changed. When I raised a
TAC case they said they had no intention of fixing it because of the
inhereant potential for causing instability. I have not bothered trying it
since then for this very reason, although it may work in newer versions of
code the potential for causing network instability makes me think it is not
worth trying.
regards
Nigel
Nigel Roy CCIE # 1405
CCSI # 21115
----- Original Message -----
From: "Joe Morabito" <joe.morabito@cox.net>
To: <>
Sent: Monday, May 27, 2002 6:20 AM
Subject: using LOAD for IGRP
> Hi guys,
>
> When forcing IGRP to use load as a metric, would you simply configure the
> metric weights command like this "metric weights 0 11100" The second
number
> "1" being the one that turns on calculating the LOAD. Or do you also need
to
> specify the bandwidth command on the interface?
>
> Thanks.
>
> 17 days and counting....
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:59:11 GMT-3