« MySQL Performance: Why Ahead Flushing for InnoDB?... | Main | MySQL Performance: InnoDB plugin-1.0.4 & others @dbSTRESS »

Tuesday, 11 August, 2009

MySQL Performance: XtraDB-6 & others @dbSTRESS

I'm happy to present you my first benchmark results with XtraDB-6 on dbSTRESS. Percona team made a huge work preparing this release and there are really a lot of improvements regarding performance as well general usage (for more details about XtraDB-6 see the full announce  from Percona site).

But my main interest is around performance (sorry :-)), and I was curious how well now XtraDB-6 resists to the stress workload. New release also integrating the "timed based" concurrency model introduced within MySQL 5.4 - missing this feature was negatively impacted XtraDB in previous tests. But now we may expect it runs at least as fast as MySQL 5.4! Let's see...

Tested versions

  • MySQL 5.4.0
  • MySQL 5.Perf (build #45)
  • XtraDB-6
  • XtraDB-6-tc (configured with "timed concurrency")

All engines were compiled from sources with GCC4.3. All binaries are using GCC atomics.

I'm not publishing a full report here yet as there was another announce today - InnoDB plugin-4 was released :-) so I wish to complete it with plugin results too :-)

But well, few graphs are more then words..

Read-Only Workload (RW=0)

Observations :

  • several times XtraDB-6 outperforms MySQL 5.4 now! - my congratulations to Percona team!!!
  • now on read-only workload the difference between concurrency settings is near to zero (Solaris 10 update7 helps a lot here!)
  • timed concurrency model is still more optimal vs original
  • good to see performance increasing when moving from 8 to 16 cores, but I'd expect more, not you? ;-)

Read+Write Workload (RW=1)

Observations :

  • limiting the concurrency is the must to keep read+write workload of more than 32 sessions
  • in several cases XtraDB is outpassing 5.4 again!
  • build45 demonstrating the best performance here
  • moving from 8 to 16 cores still improving performance, but hope one day we'll see x2 times difference! :-)

10Reads/Write Workload (RW=10)

Observations :

  • 10Reads per Write (RW=10) is less aggressive on writing and may represent a more common kind of workload for some users, and as you see limiting concurrency helping again to keep higher workloads and lower performance decreasing..
  • timed model is more optimal again
  • in some cases XtraDB performs better then 5.4, in some cases it's 5.4
  • but build 45 is still the best, seems the its main improvement was made on writing, and it confirms other our observations..

Other results are coming soon! Well done, Percona!!! My sincere congratulations! To outperform 5.4 was not an easy task! ;-))

Any comments are welcome :-)

Posted by Dimitri at 22:30
Categories: MySQL, Solaris, Tools/ dbSTRESS