Applies to: Exchange Server 2007 SP1, Exchange Server 2007
Topic Last Modified: 2008-06-23

Microsoft Exchange Server 2007 introduces service resource management to detect and take action on heavily loaded Exchange servers. When a system is under heavy load, more load should not be added. Exchange 2007 servers with either the Hub Transport or Edge Transport server roles installed have several minimum resource requirements that must be maintained. Thresholds for different resources are managed by the Microsoft Exchange Transport service (MSExchangeTransport.exe).

When an Exchange 2007 Hub Transport or Edge Transport server exhausts these monitored resources, the service will stop accepting new messages until the resources reach acceptable levels. This situation is called back pressure. For more information about back pressure, see Understanding Back Pressure.

When these resource requirement thresholds are exceeded, Event IDs 15001, 15002, or 15003 are logged. In addition, Microsoft Exchange Server 2007 Service Pack 1 (SP1) includes events 15004 and 15005. 

For all resources, a value of Normal is within normal operating levels, Medium indicates potentially high utilization, and High means that the server is resource-constrained and will stop accepting new messages. In that case, symptoms such as the following occur:

The default values and troubleshooting suggestions for each resource that is being monitored are listed in the following table. In most cases, consider running the Exchange Mail Flow Analyzer tool because this tool not only provides you with these suggestions, but it also reviews the server's overall health.

Note:
CPU and network utilization are not monitored by the Microsoft Exchange Transport service.

Troubleshooting suggestions for resources monitored by the Microsoft Exchange Transport service

Resource being monitored Description Troubleshooting suggestions Default High value Default Medium value Default Normal value

Private bytes used

The PercentagePrivateBytesUsed parameter is used to monitor the percentage of private bytes used by the EdgeTransport.exe process. The monitor is checking to make sure that the private bytes used does not exceed a default private bytes limit. For x64 computers, this limit is equal to 75% of the total physical RAM, or 1 terabyte (whichever is less).***

The EdgeTransport.exe process consumes memory as the queues fill. Check the queues to make sure that there are not any issues. If there are issues, troubleshoot the queues with the Exchange Mail Flow Analyzer (found in the Exchange Toolbox).

75%

73%

71%

Physical memory used

The PercentagePhysical MemoryUsedLimit parameter is used to monitor the total amount of memory in use by all processes.

Situations where the default High value is exceeded can be caused by queuing messages, which you probably want to troubleshoot. Often, the server attempts to reclaim memory by removing the least active messages from memory (dehydrating the queue). If this problem occurs frequently, consider removing roles from the server or re-evaluating your hardware (for example, add memory or add additional servers).

Dehydrating the queue means that unnecessary elements of queued messages are removed from cached memory, but still remain in the queue.

94%

89%

84%

Database disk space used

The intent of the PercentageDatabaseDiskSpaceUsed parameter is to monitor for available space for queued messages. This is done by monitoring the amount of used space in the database and comparing it to the amount of free space in the database and on the disk. This computation also takes into consideration the total disk size. The minimum free space is always 4GB in the release to manufacturing (RTM) version of Microsoft Exchange Server 2007 and 500MB with Microsoft Exchange 2007 SP1.

To prevent data loss, there are situations where Exchange may stop accepting mail. This may be caused by insufficient free disk resources. Check the queues for messages backing up. If the partition that holds the queue is too small, consider moving it to a drive with more space. Remove unnecessary files from the drive that holds the queue. In Exchange 2007 RTM, this issue occurs most frequently when the drive has less than 4 GB available.

In Exchange 2007 SP1, this issue occurs most frequently when the drive has less than 500 MB available. Requirements will be higher if the transport dumpster is enabled (if cluster continuous replication is used).

*

High value minus 2%

High value minus 4%

Amount of free hard disk drive space for message queue database transaction logs

The PercentageDatabase LoggingDiskSpaceUsed parameter monitors the amount of free space on the disk to make sure that the transaction logs always have enough room for committed transactions.

To prevent data loss, there are situations where Exchange may stop accepting mail. This may be caused by insufficient free disk resources If the queue database's transaction logs are on a different drive than the database, this issue indicates that the drive is too small.

**

High value minus 2%

High value minus 4%

Number of version buckets

Extensible Storage Engine (ESE) databases keep an in-memory list of modifications made to the database known as a Version Store. The VersionBuckets parameter keeps track of the number of different versions that are in memory because it is important that they are committed to disk. The size of allocated version buckets will fluctuate under normal conditions, but sizes can increase to unacceptable levels for various reasons, such as antivirus issues, problems with Jet integrity, large messages going through transport, and disk input/output (I/O) performance. If the size becomes too high, it could indicate that the Version Store has too many outstanding modifications that have yet to be committed.

Situations where the version buckets remain high are often transient and can generally be ignored. If the problem occurs frequently, it may be a good idea to verify that you have message size limits. If large messages are not the cause, consider monitoring the disk I/O performance counters to see if there is an underlying disk performance issue that could be causing the issue.

RTM: 100

SP1: 200

RTM: 60

SP1: 120

RTM:40

SP1: 80

*   RTM Limit = 100 X(totalNumberOfBytesOnDisk – 4 GB) ÷ totalNumberOfBytesOnDisk

     SP1 Limit = 100 X (totalNumberOfBytesOnDisk – 500 MB) ÷ totalNumberOfBytesOnDisk

**  Limit = (totalNumberOfBytesOnDisk – (CheckpointDepthMax × 25)) × 100 ÷ totalNumberOfBytesOnDisk

***   32-bit Exchange is not supported in production. However, for x86 computers using the /3GB switch, there is a limit of 1800 MB or 75% of physical RAM, whichever is less. For x86 computers without the /3GB switch, there is a limit of 800 MB or 75% of physical RAM, whichever is less.

In the following example of a warning event, the disk that the queue was located on was roughly approximately 8 GB in size. The amount of free space was approximately 3.6 GB, which was insufficient for the server to accept new messages safely. The problem was resolved by moving the queue database to a larger drive.

Event Type: Warning

Event Source: MSExchangeTransport

Event Category: ResourceManager

Event ID: 15002

Description:

The resource pressure is constant at High. Statistics:

Queue database and disk space ("C:\Program Files\Microsoft\Exchange Server\TransportRoles\data\Queue\mail.que") = 54% [High] [Normal=45% MediumHigh=47% High=49%]

Queue database logging disk space ("C:\Program Files\Microsoft\Exchange Server\TransportRoles\data\Queue\") = 54% [Normal] [Normal=89% MediumHigh=91% High=93%]

Version buckets = 0 [Normal] [Normal=40 MediumHigh=60 High=100]

Private bytes = 10% [Normal] [Normal=71% MediumHigh=73% High=75%]

Physical memory load = 52% [limit is 94% to start dehydrating messages.]

For More Information

For more information about troubleshooting transport and mailflow issues, see Transport and Mailflow Issues.

For more information about the Exchange Mail Flow Analyzer, see Using the Exchange Management Console and Using the Toolbox.