Home > Workload Solutions > SQL Server > White Papers > Dell PowerMax 2500 and 8500 Best Practices for Mission Critical SQL Server Databases > Performance results with storage compression enabled
The performance test with three simultaneous HammerDB workloads is repeated with PowerMax compression enabled. PowerMax compression is enabled only for the database files SGs (not the transaction log).
The test was run twice, and the results from the second run were collected. The first run allows the PowerMax cache and data placement algorithms opportunity to learn how the system is used and compress the data. In customer production environments, system use does not change often and tends to have repeating patterns. During the first run, the PowerMax EM (the emulation responsible for data reduction and movement) was at 75 percent use, and in the second run it was at 59 percent use, which is 4 percent higher use than it was before enabling compression.
The following figure shows the test results and differences. The Transaction per Minute (TPM) collected from HammerDB shows an aggregated value of 1,627,659 TPM, which is only 1.3 percent lower than the noncompression test TPM results. The results show 177,838 total IOPS collected from PowerMax, which is 2.7 percent less than the noncompression run. The difference in read and write latency to the database files (measured from the PowerMax storage group performance metrics) shows a difference of 8 microseconds from the noncompression run, which is negligible.