12

I've been tasked to stress test our MSSQL Server and MySQL Server. I would like to know if there are any tools or scripts which I could use on our current systems and the new system to compare performance?

I would like to measure read / write to disk and processor performance. Anything else that may be useful would be great as well.

Thanks!

stanleykylee
  • 474
  • 1
  • 6
  • 21
  • 6
    For SQL Server, I'm frequently mentioning the RML tools which are discussed in http://dba.stackexchange.com/questions/5809/testing-stored-procedure-scalability/5839#5839. – Mark Storey-Smith Oct 20 '11 at 17:32
  • Thanks for the post. Great set of tools for MSSQL. I'll have to give it a shot and see what I can come up with. This actually helps in our environment as I can re-create user activity with our database.

    That said, I also found SQLIOSim from Microsoft. It doesn't quite test the DB but it does test the IO of the system which the DB lives on.

    http://support.microsoft.com/kb/231619

    – stanleykylee Oct 20 '11 at 18:23
  • 2
    Careful with SQLIOSIM, it's intended to test IO stability rather than capability http://blogs.msdn.com/b/psssql/archive/2008/04/05/sqliosim-is-not-an-i-o-performance-tuning-tool.aspx. SQLIO is the tool you want instead. – Mark Storey-Smith Oct 20 '11 at 18:34
  • Please also look at our DTM DB Stress tool (www.sqledit.com/stress). It was designed for exact requested purpose. – Igor Shekalev Jan 13 '13 at 07:02
  • JMeter comes to mind, should work for both (you could even run the same/identical if you design them properly) –  Jan 13 '13 at 12:30

3 Answers3

10

Quest offers a free version of their Benchmark Factory that works for Oracle, SQL Server, and MySQL:

http://www.quest.com/benchmark-factory/

I recorded a video on getting started with it back when I worked for Quest:

http://www.toadworld.com/Videos/tabid/831/TID/559/cid/161/Default.aspx

You can call your own stored procs, replay traces, run TPC benchmark scripts, and more. It's not exactly easy to use, though - thus the video.

Brent Ozar
  • 42,952
  • 47
  • 220
  • 375
4

For MySQL, there's the official benchmark suite. Alternatively a tool like sysbench. I'm currently in the process of testing out sysbench, but I've been having some trouble!

Derek Downey
  • 23,440
  • 11
  • 78
  • 104
2

The defacto standard for comparing one database solution to another is the TPC-C benchmark. The latest definition of this benchmark can be downloaded from http://www.tpc.org/tpcc/default.asp along with existing benchmark result sets.

While it is unlikely you will run the full TPC-C at your location, the test definition should provide the foundation for developing a set of internal benchmarks designed to identify performance differences on a reference set of hardware with respect to your organization's specific requirements for transactional throughput.

Make sure you download the cost of ownership documentation as well, for this can provide a standardized method for evaluating the two solutions.

James Pulley
  • 343
  • 2
  • 4