RE: Summarised Backup Network problem

From: Scott Vermillion (scott_ccie_list@it-ag.com)
Date: Wed Mar 05 2008 - 17:16:47 ARST


And just to clarify a bit here...

I'm not advocating both of the below solutions. I think the L3 connectivity
between your routers would seem (on the surface at least) to solve most if
not all of your issues. But if you can't or don't want to do that for
whatever reason, then consider the static route/track object/redistribution
concept to control whether or not the primary router advertises the summary
towards the datacenter...

-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
Scott Vermillion
Sent: Wednesday, March 05, 2008 12:05 PM
To: 'Ryan Morris'; ccielab@groupstudy.com
Subject: RE: Summarised Backup Network problem

Hey Ryan,

Any chance of establishing a direct L3 connection (not via the switches)
between your two routers? Doesn't seem like you'd *want* to take your
backup WAN circuit (unless they're equal in terms of capacity and
performance) if the primary WAN circuit was still alive, which presumably it
still would be in one of these scenarios where you just lose the switch tied
to the primary router/circuit). If you have the router interfaces
available, that would seem like the most obvious solution to me, personally.
In terms of how to control the advertisement of your summary towards the
datacenter, there are probably a couple of ways to deal with that (e.g.
create a static route to null0 for your summary range with a track object
for the attached switch's loopback and then redistribute this static into
your IGP (or not) based upon that reachability (or not)). Something like
that?

Regards,

Scott

-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of Ryan
Morris
Sent: Wednesday, March 05, 2008 11:16 AM
To: ccielab@groupstudy.com
Subject: Summarised Backup Network problem

Here's a scenario I've run into in real life:

We have a branch office with two WAN connections, primary and backup.
Traffic will only take the backup link if the primary is not available.
We run EIGRP between these routers and our data centre routers. I'm
planning to summarise the routes coming out of these branch routers in
order to simplify my routing table. Per best practice, there is a
loopback address in each of the branch routers that is in the netblock
for that office.

Inside the branch office, there is a group of core switches made up of two
3550s. Each 3550 connects to one of the WAN routers, and has an EIGRP
relationship with the other 3550 and the connected router.

So if the primary WAN link or the primary router fails, no problem.
Traffic routes to the backup.

Problem: if the connection between the primary router and the 3550 fails
(or, let's say the switch dies), that router will continue to advertise
the summary because of the loopback, and because it has a better
metric than the backup, traffic will not fail over to the backup.

Any ideas on how to solve this? i.e. a feature that shuts down an
interface or explicitly stops advertising a route if another interface
fails? Or is the the simple answer (take the loopback off the primary
router) the only way to keep this from happening?

Input appreciated!

Ryan Morris
CCIE #18953



This archive was generated by hypermail 2.1.4 : Tue Apr 01 2008 - 07:53:52 ART