Hi,
first the easy case.
If you terminate dlsw peers on the cip router itself it would look like this:
Based on your example, from the remote branch you configure a dlsw peer to each of the two cip routers.
Dlsw by default talks to all source-bridg ring-groups you have configured on a router and thus the remote router would learn via both peers the reachability for 4000.0755.0001. Since your source-bridge your local tokenring lans on the cip into the source-bridge ring-group.
Typically customers configure on the remote branch router a cost on the remote peers. I.e. cost 2 on peer 1 and cost4 on peer 2. For half the remote peers you do it this way and the other half you flip the costs , that way you achive load balancing.
that means if one peer, cip router, or the xca fails than the sessions over this connection will fail and can be reestablished without any user intervention over the second connection, provided that it is up and running and can reach the needed host applications.
Now for the ethernet attached clients the solution is not that simple.
The problem here is that you can not have the same mac address twice in an ethernet segment/vlan.
Solution 1:
The simple solution is to split the ethernet and connect half of the clients to the first vlan connected to ciprouter1 and connect the other half to a second, separate, vlan on ciprouter2. You configure sr/tlb on both of them for the bridging between the ethernet vlan and the cip. Thus each vlan can see the host mac address always only once.
However if one cip router fails half of your local ethernet clients are offline.
Solution 2:
use snasw on the cip routers. This requires that you have appn configured on the host. It also requires a snasw license on the router.
You can then configure the local ethernet as a snasw port and use hsrp on the ethernet to communicate between the two routers. hsrp can also be used to create the necessary mac addresses you have on the host with the hsrp mac-address command.
If you need more information about snasw, please have a look at cco. There is quite some documentation out there including design guides ect.
Solution 3:
We have a special feature on current cip microcode called HSMA. Please dont confuse this with hsrp. it is similar but not the same.
HSMA provides a mechanism for two cips on two different routers to talk to each other and present always only one mac address to the ethernet segment you are connected to. When the active cip fails the standby one comes active and it presents the same mac address to the network. So that your end systems can reconnect without manual intervention.
If you need more information about HSMA you need to send me a private email so that i can forward you some detailed documentation. It is not yet out on cco.
My email address is mbinzer@cisco.com
thanks...
Matthias
