Our DR / Continuity plan includes Exchange 2003 clusters in synch with backup clusters in different cities. An event requiring a "site failover" requires DNS updates so we can bring up the backup cluster with the same name as the production cluster. While we are migrating users to these clusters, we still have Exchange servers in the same routing group as these clusters that must be prevented from delivering mail to these clusters for about 12 hours while we conduct a failover / failback test.
All of our servers are in a single routing group so that, so that they can easily route around any points-of-failure.
I've considered changing the port that SMTP uses on the clusters for this duration so that all of the other Exchange virtual servers.