Considering Other Factors That Affect Failover
The failover options that you set for your Exchange Virtual Servers are only one factor that affects the speed at which an Exchange Server 2003 cluster fails over. In addition to those settings, many other factors can influence failover rates. The following table lists these additional factors. By understanding these factors, you can configure your Exchange clusters for optimal failover. For detailed instructions, see How to Add the MsgHandleThreshold Registry Key Value.
Factors that affect failover performance of Exchange 2003 clusters
| Factor | Description |
|---|---|
State of the Exchange store |
The state of the Exchange database and logs at the time of startup or shutdown affects failover performance. For example, if Exchange databases were shut down abruptly, there may be lots of log files to roll through before starting the Exchange databases on the new Exchange Virtual Server. |
Number of storage groups and databases on your servers |
Generally, the greater the number of Exchange databases on your Exchange Virtual Server, the longer it takes to move resources to the new Exchange Virtual Server. |
Number of service connections into the Exchange store |
The Exchange store performs cleanup routines before it releases and allows failover to occur. An unloaded server that takes 100 seconds to fail over takes 120 seconds to fail over when that server has 3,000 simultaneous Microsoft Office Outlook® Web Access or Microsoft Outlook connections. |
Size of the SMTP queue |
If the SMTP queue size is greater than 1,000 messages, the time to fail over from one cluster node to another can be significant. You can modify this setting by creating and configuring the SMTP Max Handle Threshold registry key value:
For more information about creating and configuring this registry key, see the procedure following this table. |