Is it possible Bandwidth on Demand over 2 links with OSPF?

[sisal]

I have 2 sites connected with one primary link and one acting as backup. I've defined the backup line rising its OSPF cost instead of setting "maximum-paths" to 1 (and leaving costs equal) because I want to have always the second line as backup.

Can I activate the backup line in case of demand for further bandwidth? And how?

I know that OSPF balances the load only over equal cost lines but I would like to keep the backup line DOWN when it's not needed.

I've thought of CEF, but I dont know how to implement it to obtain Bandwidth on Demand. In the case that CEF is not appropriated, should I implement policy-based routing using a Route Map to direct the extra traffic to the backup line when needed? Any suggestion is appreciated...

Thanks, Piero

[rais]

I think you can benefit from the following commands:

interface serial 0

backup interface serial 1

backup load 85 5

backup delay 10 60

Ser1 is backup for ser0 (could be any type of interface). In case ser0 remains down for 10s, ser1 comes up. In addition, when load on ser0 reaches 85%, ser1 is activated.

I have never tried it, but believe OSPF on backup links will be redundant in this situation. I think you should have OSPF with max-path = 2 and equal costs. Adjacencies on backup-link will be formed only after backup comes up.

HTH

[sisal]

I use "backup" as a feature of the LAN interfaces on the HQ router.

As far as WAN is concerned, I didn't specify that encapsulation of the 2 links is frame-relay (the backup line via 2 Aethra SHDSL FastFrame boxes) , so if the PVC is "deleted" the loss of keepalives on the local interface activates immediately the backup line, but I'm afraid that if the PVC is "inactive", that is locally the protocol is UP (the problem is on the remote node) the backup interface NEVER comes up. I'm I right?

Piero

[rais]

Even if PVC goes inactive, backup will kick in for that particular sub interface. You can use backup under sub-interface as well.

Thanks.

[sisal]

Just checked, you're right; shutting the local interface, protocol went down also on the remote sub-interface. I'll have to review my knowledge of frame-relay protocols, LMI, keepalives etc..

So: your suggestion works, is simple to activate, even avoids OSPF hello traffic that flows on the backup line...yet, are you shure that is the best (or only) way to obtain Bandwidth on Demand over those 2 links?

Thanks again

[rais · Accepted solution]

As far as I know, this is one of the solutions to accomplish this task. However, there may be other ways known to others.

Thanks.

[ruwhite]

This is the only way I know of, at present. You could possibly use GLBP, with no routing protocol, but if you can do it with backup interface, that's probably going to be better, I'd think, since you can preserve routing.

The reason the remote subinterface goes down is because the PVC goes down, so the A bit is cleared on the PVC, and the router brings the subinterface down. If you were running this as a multipoint, that wouldn't work. Make certain, if possible, that you're running asynchronous LMI, which will give you faster down notification.

Russ.W