Calculating the verification throttling sleep interval

You must calculate the appropriate sleep interval value to use for each server. Monitor the read and write performance of the LUNs on which the databases reside, and then calculate the verification throttle setting so that the average and peak physical disk performance is below the maximum values determined by Microsoft.

About this task

Steps

1. Launch the Windows Performance Monitor utility (also known as “Perfmon”) on the machine that is the Exchange server.
2. Add the Perfmon counters PhysicalDisk\Average Disk sec/Read, and PhysicalDisk\Average Disk sec/Write for each LUN that contains a database.
3. Monitor the pairs of values, Physical Disk\Average Disk sec/Read and PhysicalDisk\Average Disk sec/Write, while a SnapManager verification job or a backup job with verification is in progress.

   If monitoring a backup with verification job, be sure to monitor the counters during the verification portion of the job. Multiple databases on the same Exchange server are verified sequentially, so be sure to monitor the counters for all of the database verifications.

4. The throttle setting is global, so select the reading from the database that exhibits the worst performance (highest readings).
5. Compare the measured values to the maximum read and write values recommended by Microsoft.

   Physical Disk\Average Disk sec/Read: Average read time should be below 20 ms, Maximum peak read time should be below 50 ms.

   Physical Disk\Average Disk sec/Write: The maximum physical disk performance times determined by Microsoft are as follows: Average write time should be below 20 ms, Maximum peak write time should be below 50 ms.

   Compare the measured values to the maximum read and write values recommended by Microsoft: less than 20 ms. average time and less than 50 ms. peak time.

6. If the measured values exceed the maximum values recommended by Microsoft, adjust the throttle setting to decrease the verification throughput rate.