Monitoring for Exchange 2000 memory fragmentation (296073)

SUMMARY

The Exchange scalability and performance test team has identified a clustering scenario in which an active node is unable to reliably take over the Exchange 2000 workload from another active node. There are two new ways to monitor virtual memory (VM) fragmentation after the installation of Exchange 2000 Service Pack 1 that help prevent such a scenario from occurring. New System Monitor counters are noted in the following table.

Performance Object  |   Performance Counter              |  Description 
-------------------------------------------------------------------------------------------------------------------------------------
MSExchangeIS        |   VM Largest Block Size            |  Size in bytes of the largest free virtual memory block.
                    |                                    |
MSExchangeIS        |   VM Total Free Blocks             |  Total number of free virtual memory blocks.
                    |                                    |
MSExchangeIS        |   VM Total 16MB Free Blocks        |  Total number of free virtual memory blocks larger than or equal to 16 MB.
                    |                                    |
MSExchangeIS        |   VM Total 16MB Free Block Bytes   |  Total number of bytes in free virtual memory blocks larger than or equal
                                                         |  to 16 MB.

Note These counters will not work on a passive node of an active/passive cluster. This is because the Microsoft Exchange information store service (Store.exe) does not run on the passive node of the cluster. To add virtual memory counters for the information store, at least one virtual server must be started on the cluster node. Also, these warning and error messages are logged in the Application event log when certain levels of VM fragmentation are reached. When the server reaches less than 32 MBs of contiguous virtual address space, the following event ID 9582 warning message is logged:

Event Source: MSExchangeIS
Event Category: Performance
Event ID: 9582
Type: Warning Description: The virtual memory necessary to run your Exchange server is fragmented in such a way that performance may be affected. It is highly recommended that you restart all Exchange services to correct this issue.

For more information, click http://search.support.microsoft.com/search/?adv=1.

When the server reaches less than 16 MBs of contiguous virtual address space, this same event ID 9582 message is logged, but it is logged as an error message:

Event Source: MSExchangeIS
Event Category: Performance
Event ID: 9582
Type: Error
Description: The virtual memory necessary to run your Exchange server is fragmented in such a way that normal operation may begin to fail. It is highly recommended that you restart all Exchange services to correct this issue.

For more information, click http://search.support.microsoft.com/search/?adv=1.

MORE INFORMATION

The event ID 9582 that is referenced in this article is primarily a concern for Active/Active cluster configurations. In Active/Passive cluster configurations and in non-clustered Exchange 2000 computers, event ID 9582 can be ignored unless server performance is affected. A feature that modifies this behavior is included with Exchange 2000 Server Service Pack 1.

For additional information about the latest service pack for Microsoft Exchange 2000 Server, click the following article number to view the article in the Microsoft Knowledge Base:

301378 How to obtain the latest Exchange 2000 Server service pack

The "VM Largest Block Size" counter that is mentioned in the "Summary" section of this article indicates how much contiguous memory is available. When the contiguous memory falls below 16 MB, the information store begins to log the event ID 9582 (error) messages. This event ID may be a concern only if you want to mount or dismount a database. That operation requires at least one 10-MB contiguous section of Virtual Memory to replay log files, and if the virtual memory is not available, the operation cannot mount the database. A restart of the Information Store service may permit the store to be mounted successfully.