Improving MySQL performance by up to 8x using Adaptec MaxIQ™ SSD Cache Kit with DataON DNS-1400 JBOD System.

People want the database blocks with high cache localilty to be in the SSD and everthing elase can remain on the voluminous, hight capacity, low price-per-byte disks, the goal of this engagement is to capture the performance statistics on the Adaptec RAID 5445Z controller with MaxIQ™ SSD Cache disabled and DataON Pay-AS-Your-Grow JBOD Storage system DNS-1400  for a comparative analysis.

Adaptec MaxIQ™ SSD Cache Performance Solution for MySQL Analysis : Download PDF

Test Objectives

Tests were conducted with the following requirements for the MySQL Server environment:

• MySQL Server with 82,000 records hosted on 1 MySQL server.
• Obtain performance metrics with MaxIQ SSD cache on the Adaptec 5445Z host bus adapter disabled, followed by identical testing with MaxIQ SSD cache enabled which include MySQL server statistics and other server operating system performance metrics.
• Executed each test run for 60 minutes.

Software

The Following table describes the software that was used for testing the MySQL Server environment:

Hardware

The following table describes the hardware that was used for testing the MySQL Server environment:

Analysis Result:

These are the test results of the Adaptec MaxIQ SSD Cache Performance Solution with MaxIQ cache enabled compared to being disabled. In all cases the MySQL Server was monitored to ensure the only bottleneck was the storage controller. CPU, RAM, network, MySQL and other devices were monitored to ensure that they were not causing the environment to perform slowly.

Throughput with 95% Reads and 5%Writes back with:

• When MaxIQ SSD cache disabled the MySQL throughput stabilized at 67.6MB/s, but MaxIQ SSD cache enabled the MySQL throughput stabilized at 542.3MB/s;  performance increase of 8.0x with cache over no cache.
• When MaxIQ SSD cache disabled the MySQL transactions stabilized at 346 transctions, but MaxIQ SSD cache enabled the MySQL transactions stabilized at 2374 transactions;  performance increase of 6.9x with cache over no cache.

Conclusion

The following items provide an overview of the tests comparing the Adaptec MaxIQ SSD Cache Performance Solution within a MySQL Server environment:

• Read/Write test run with 95% select statements and 5% insert/delete/update statements.
  • Throughput in mega bytes per second (Mbps) went from 67.6 Mbps with SSD cache disabled to 542.3 Mbps with SSD cache enabled for a performance gain of 8.0 times the non-cached environment.
  • Transactions per second went from 346 transactions per second with SSD cache disabled to 2374 transactions per second with SSD cache enabled for a performance gain of 6.9 times the noncached environment.
• The test environment was designed so that CPU utilization, network traffic and memory were not bottlenecks, and disk I/O through the Adaptec RAID 5445Z was heavily loaded to exercise the 5445Z’s cache.

Installing, configuring and managing the controller and SSD cache is a trivial, quick and intuitive experience.Managing the MaxIQ cache only requires half a dozen mouse clicks from launching to exiting the GUI interface.