Is there any correlation between the number of users on an Exchange cluster and the amount of time it takes to fail over from one node to the other?
There isn't a correlation. In an Exchange 2000 Server cluster, you create a virtual Exchange server that users connect to. Their Microsoft Outlook or Internet-protocol clients establish connections to the virtual server, and the cluster-management software routes those connections to the appropriate physical node. For example, you might have a virtual server named EXCHANGE pointing to physical machines named SCULLY and MULDER. Users' profiles point to EXCHANGE. If my mailbox is on MULDER, when that machine fails over, the Exchange-specific cluster resourcemanagement DLLs fail the mailbox store over to SCULLY, and I can immediately reconnect. Failover time is a function of the time it takes to complete the physical transfer of the quorum disk and shared storage, plus the amount of time it takes to start the failed-over services on the target node, including the time that it takes to play back transaction logs, contact a Global Catalog (GC) server, and perform the other tasks required during a regular startup.