Using Jetstress to Verify Storage System Performance

 

Now that you've evaluated your I/O requirements and optimized your disk subsystem according to the best practices for your specific storage architecture, you should verify that your disk performance will meet your expectations prior to configuring it in a production environment. To do this, you can use the Jetstress tool against your storage system in a test lab.

To download Jetstress, including comprehensive installation and usage instructions, see the Downloads for Exchange 2003 Web site.

Jetstress is a tool that helps verify disk performance by simulating Exchange disk I/O load. Jetstress simulates the Exchange database and log file loads produced by a specified number of users. You use System Monitor, Event Viewer, and Exchange Server Database Utilities together with Jetstress to verify that your disk subsystem meets or exceeds the performance criteria you establish (minimum acceptable latency while under your IOPS per mailbox load).

With Jetstress, you can perform two types of tests:

  • The Jetstress Disk Performance Test runs for two hours. This test allows you to verify the performance and sizing of your storage solution and gives you the saturation limit for your drive. You can use this information to verify that your storage system will provide you with acceptable performance when fully loaded.

  • The Jetstress Disk Subsystem Stress Test runs for 24 hours. This test allows you to stress your server by using a much larger load over a more significant amount of time. It allows you to find out if your storage system will perform erratically when under excessive load.

Running both tests is the best way to verify the performance of your disk subsystem.

After successful completion of these tests in a non-production environment, you are ready to implement your storage solution into your production environment. You should never run Jetstress on a live production server.

Note

In addition to Jetstress, Load Simulator 2003 (LoadSim) is another tool that is commonly used to evaluate storage system performance. However, LoadSim does not account for all of the factors involved in sizing servers. As a result, LoadSim does not provide a comprehensive representation of user experience—the results should not be interpreted in that aspect. When evaluating performance of your storage system, you should use Jetstress.