Skip to content

Loading…

API Performance improvements on very large setup: SQL query tweak #3140

Closed
mattab opened this Issue · 4 comments

2 participants

@mattab
Piwik Open Source Analytics member

See in forum

We could change the archive select sql query to make it faster and use an index instead of memory sort.

@diosmosis
Piwik Open Source Analytics member

Attachment: Benchmarking patch.
3140.diff.tar.gz

@diosmosis
Piwik Open Source Analytics member

Been looking into this issue. I created a benchmarking system to test the changes I made. However, there doesn't seem to be any change in speed when applying the changes SQL changes specified in the forum... Maybe I'm not adding enough data?

@mattab
Piwik Open Source Analytics member

Nice that you investigate this one, it's not easy to reproduce such issues indeed.

  • Did you try on mysql 5.5.x ?
  • in the bug report the user had about 1M rows in the tables, so maybe the bug shows up with more?
  • Also you can try changing the tables to InnoDB which might be one factor playing a role in this regression?
@mattab
Piwik Open Source Analytics member

(In [6402]) Fixes #3140 Can it really be that simple??

@mattab mattab added this to the Piwik 1.8 milestone
This issue was closed.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Something went wrong with that request. Please try again.