People ask about creating redundancy for thier VSM servers. There are a few redundancy options available for VSM, but one option we've had some questions about lately is NIC teaming.
The NIC teaming or bonding feature is not officially supported by Cisco for the VSM platform. We’ve had people try it in the past and the biggest problem they encountered was with performance. If you have packets from a given stream being sent along two different paths it introduces a timing issue where the traffic may not arrive in the right order.
The approach that we have seen work is to have the second NIC given its own address. Having the second NIC use its own address allows you have the redundancy you’re looking for. In order for things to work correctly though, you'll want to have the DNS name of the media server configured in VSOM instead of the ip address. That would give you some control over the address clients are being sent to. If you know that the main address will be down for example you could just change the settings in DNS and everyone would start going to second address instead (even client workstations).
If the main IP address stops working any proxies running across that interface would be disconnected, but the proxy process will initiate outbound connections over the remaining interface shortly thereafter.
As with any contingeny plan, one would want to make sure you test the solution out to make sure it works the way you expect it to.
I'm curious what other people have tried in this area. What have you seen work?
Thanks for this Gerry. One of the symptoms I remember seeing caused by bonding two NICs was that the mrtg process would not start, leaving the server in a "Partial" state as shown in the Servers page within VSOM.