From: Collins Bill (parfour_on_9@xxxxxxxxxxx)
Date: Sat Mar 04 2000 - 10:35:58 GMT-3
Kevin,
I can answer number one. I did not fully understand question 2.
Split horizon is always an issue with NBMA. With distance vector protocols,
split horizon is disabled on physical IP interfaces whereas with
subinterfaces split horizon is enabled. This is very important.
With EIGRP, one of the benefits is when you route IPX using EIGRP. With IPX
RIP, there is no way to disable split horizon. So this means that in order
to properly route IPX over an NMBA network, turn on EIGRP for IPX. Enabling
EIGRP gives you the ability to disable split horizon at the interface level.
Bope this helps,
Bob
>From: "Kevin Gannon" <kevin@gannons.net>
>Reply-To: "Kevin Gannon" <kevin@gannons.net>
>To: <ccielab@groupstudy.com>
>Subject: EIGRP ?
>Date: Sat, 4 Mar 2000 13:06:23 -0000
>
>A couple of EIGRP questions :
>
>1. Why is it that you must explicity disbale split horizon for EIGRP on
>frame-realy (yet not for IGRP/RIP) is there
> some logic on Cisco's part can it badly break something if you turn it
>off ?
>
>2. If you want to do unequal load balancing over two path using yet the
>other paths which are also unequal you want
> to leave as they are , can this be achieved with the variance command
>or must you manipulat the metric inwards
> from these paths using route maps on a per neighbour basis rather than
>using the metrics statement which affects
> all the neighbour updates ?
>
>Regards,
>Kevin
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 08:23:03 GMT-3