DLSw - How to route remote site traffic to HQ via a Regional Site?

[m.laporta]

Hi Experts.

My Customer is a Bank with a two-tiered WAN: 100 small remote sites directly connected (leased lines) to HQ.

There is also a Medium Regional Site with redundant WAN connection to the HQ. Remote sites and Regional Site have a DLSw peering to the HQ.

Now the Customer needs to migrate to a redundant, ISDN-based connectivity solution for remote sites.

To keep the solution cheap, it is best to make the remote site call the Medium Regional Site.

In other words, the backup network is three-tiered:

HQ---RegionalSite---RemoteSite.

What's the best practice to configure the Regional Site to forward DLSw to the HQ in backup conditions?

Will this work config work:

- regional sites have a DLSw backup peering with Regional Site

- Regional Site is a DLSw border peer for regional sites and has a standard DLSw peering with HQ

Thank you in advance

michele

[drolemc]

You might want to take a look at this document that describes how to backup dlsw

http://www.cisco.com/univercd/cc/td/doc/cisintwk/idg4/nd2007.htm#xtocid26

[dmccullo]

Hi Michele,

The document Donald provided covers it well from the DLSw+ perspective. What may be missed in the scenario you describe, is that you don't need to terminate the DLSw+ peers at the regional site, only establish an IP route through the regional site, back to HQ. So the DLSw+ peer connection is between the remote and HQ, in either the primary or backup state. The key is Dial-On-Demand Routing (DDR), and getting the IP route re-established -

http://www.cisco.com/en/US/tech/tk801/tk133/tech_protocol_family_home.html

Rgds, Dan

[jvogler]

Create a peer statement to the main HQ on the remote sites. Then create a backup peer with a linger statement to prevent a link flap wreaking havoc on the remote sides.

E.G.

HQ loopback/peer address address 1.1.1.1

Regional site loopback/peer address 2.2.2.2

dlsw remote-peer 0 tcp 1.1.1.1

dlsw remote-peer 0 tcp 2.2.2.2 backup-peer 1.1.1.1 linger 10

This will allow the remote site to use the HQ if he looses the peer at the HQ and it will wait until the HQ is up for 10 minutes before moving the peer back to the main site. Make sure you use the linger command so your peers don't end up where you don't want them.