Topic Last Modified: 2007-01-23

The Microsoft® Exchange Server Analyzer Tool uses the Exchange Server User Monitor (ExMon) tool to determine whether user MAPI Online operations are taking longer than should reasonably be expected on a healthy Exchange server.

An example of a MAPI online operation is the QueryRows operation.

The QueryRows operation is used to return rows in the current view. QueryRows uses the data in a table to satisfy the user request. The SetColumns operation determines the data available to QueryRows. The cost of QueryRows increases with the size of the properties requested, the number of properties requested, and the number of rows of data requested. QueryRows can also be expensive if many columns and rows are requested.

As the number of items in the core Exchange Server 2003 folders increase, the physical disk cost to perform some operations will also increase for users of Outlook. Sorting the Inbox by size for the first time is very expensive (requires many disk I/Os). Future sorts of the Inbox by size will be very inexpensive.

If the Exchange Server Analyzer determines that, over the timeframe data was collected, a user has had one or more MAPI Online operations take longer than 30 seconds, the Exchange Server Analyzer displays an error.

Long latency in MAPI Online Operations can indicate a server resource bottleneck. For the QueryRows operation, it may also indicate a high-item count in a folder or many properties set on the internal tables.

If the high-latency Online operation is operating on lots of data at the same time, the high latency may not indicate a problem with server performance. An example of this is when a user selects all the items in a folder and marks them as read in a single operation. Work with the user experiencing the high latency to determine:

To address this issue:

Some applications can significantly increase server load without issuing lots of MAPI Online operations. This is because some operations are more expensive than others. It may take only a small increase in the number of costly operations to noticeably affect server performance. In ExMon, these users are reported as having a relatively high CPU impact, without necessarily having issued lots of MAPI operations.

Also be aware that whenever there is a bottleneck for resources (generally a disk or CPU bottleneck), the latencies for the MAPI Online operations will increase. Identify server resource bottlenecks and either increase server resource capacity or move users to a less-loaded server.

For More Information

For more information, see the following Exchange Server resources: