DLSW Backup peers

[hbca]

How many backup peers can I define? How can I manage a priority between them?

In our network we have for routers - two for the Primary Data Center and two for the DR. Each remote site should have only one active peer at the time. We don't need load balancing. If first router at the Primary DC goes down I want to have peering to the second router at the Primary Data Center. If Primary Data Center goes down I want to have peering to the first router at the DR site and so on. I am wondering if I can assign priorities for the backup peers and how?

[jolmo]

- I think you can define one backup peer for each remote peer.

- The "dlsw remote-peer" command has a 'priority' option:

http://www.cisco.com/en/US/products/sw/iosswrel/ps5207/products_command_reference_chapter09186a00801a7e59.html#1050836

For your scenario configuration this might help:

http://www.cisco.com/en/US/tech/tk331/tk336/technologies_tech_note09186a0080094548.shtml#sol3

You can also see:

http://www.cisco.com/en/US/tech/tk331/tk336/technologies_configuration_example09186a0080093ecb.shtml

HTH

[hbca]

Thanks for the information and links.

[hbca]

Thanks for the information and links.

[mbinzer]

Dlsw backup peers are defined at the branch router. Each remote peer can have exactly one backup peer. There is no way to define a order or priority between backup peers.

A backup peer is a backup when the tcp connection of the primary peer gets lost. In that case the backup peer comes active.

It backs up the tcp connection between the two routers forming the primary dlsw peer. Nothing else. If you have an error somewhere on the data center behind the dlsw peer a backup peer does nothing for you.

If you have 4 head end routers than i would advice to make the two for the Primary data center active and let the branch routers decide over which one the actual circuits go. If you use the DR only as DR you can manualy disable dlsw on those locations until you actualy need it.

dlsw disable

Alternative is to have the mainframe, XCA's ect not active in the DR. And have all remote sites have four active peers.

By doing it that way you will always have a connection no matter what failure you get and the switchover can occur with no or minimum user intervention. ( e.g. the XCA's have to be manuly activated )