Pete Welcher at Chesapeake NetCraftsmen recently told the interesting story of a customer who added a backup data center. Their network was configured to support multicast. Pete describes the problem very succinctly:
If you have a customer doing multicast, and they add a 2nd Data Center or DR Site at L2, they’ll probably give the new router/switch the next higher IP on each VLAN. The problem with that is that it shifts the PIM DR from the main campus to the DR Site. It may be somewhat invisible and work, unless they don’t enable IPmc on say the links back to the main campus core from the DR Data Center.
In the case I just ran into, there was a new DR switch with an IP on exactly one VLAN, PIM enabled on the VLAN but no other interfaces. It took a while to figure out why the main campus Data Center server switch wasn’t sending PIM Registers for the one IPmc server on that VLAN. The answer turned out to be, the other one had the higher IP, it was PIM DR, so it was trying to do the Registers but did not have PIM enabled on the links on the dual unicast path to the RP…
What Pete describes is one of the reasons that I don’t like to do L2 across the WAN. There are numerous failures that can occur and such failures are often very difficult to troubleshoot, especially with a transparent bridged L2 infrastructure.
Re-posted with Permission
NetCraftsmen would like to acknowledge Infoblox for their permission to re-post this article which originally appeared in the Applied Infrastructure blog under http://www.infoblox.com/en/communities/blogs.html