From: nasir.shaikh@atosorigin.com
Date: Tue Mar 08 2005 - 03:51:44 GMT-3
Hi guys,
This is NOT something that I have come across in any practice lab scenario but am quite curious about the possibilities
Suppose we have the following situation:
12 34 ss ss
\\ // \\ //
__\\_//__ __\\_//__
|___r1___|----|__r2___|
| |
___ |____ ___|____
|___s1___|----|__s2___|
R1 has WAN connections to 2 remote locations. Links 1 and 2 do load balancing (per-destination) to destination A and links 2 and 3 to destination B.
R2 has shadow links to the same destinations. Under normal circumstances, the shadow links are not used. Let's say link 1 on R1 goes down thus reducing the bandwidth to dest A by 50%. In such a situation I want the traffic to use only the shadow links on r2 for destination A. The "bundle" of lins 1 and 2 should not be used at all. For destination B the links on R1 should continue to be used.
All devices have layer 3 connectivity. The S1 and S2 are layer 3 capabale switches (e.g. 3550 EMI,). R1 and R2 are say 7200 routers. EIGRP is used on all devices. The remote site has a similar setup.
Is there a technology or feature that can be used to make the network consider the bundle (1 and 2) as unusable and switch over to the shadow links? Using PPP reliable-link is an option, I think but would give a lot of overhead if the WAN links are ATM pvcs.
Any ideas?
Nasir
This archive was generated by hypermail 2.1.4 : Sun Apr 03 2005 - 17:56:42 GMT-3