sysbench-developers team mailing list archive

Thread
Date

[Question #128209]: Making ROWS_BEFORE_COMMIT different from INSERT_ROWS biases the test toward MySQL

To: sysbench-developers@xxxxxxxxxxxxxxxxxxx
From: Alexey Kopytov <question128209@xxxxxxxxxxxxxxxxxxxxx>
Date: Wed, 06 Oct 2010 21:32:32 -0000
Reply-to: question128209@xxxxxxxxxxxxxxxxxxxxx
Sender: bounces@xxxxxxxxxxxxx

New question #128209 on sysbench:
https://answers.launchpad.net/sysbench/+question/128209

I noticed in the source that there was a difference between ROWS_BEFORE_COMMIT and INSERT_ROWS, which biases the test against PostgreSQL, which wants to do as few commits as possible.  Is there some way either to make them the same number, or to make the parameter only apply to systems where it can actually help?

dfetter-dev1:dfetter:~/Downloads/sysbench-0.4.12 ack ROWS_BEFORE_COMMIT 
sysbench/tests/oltp/sb_oltp.c
32:#define ROWS_BEFORE_COMMIT 1000
473:      if (commit_cntr >= ROWS_BEFORE_COMMIT)
480:        commit_cntr -= ROWS_BEFORE_COMMIT;

dfetter-dev1:dfetter:~/Downloads/sysbench-0.4.12 ack INSERT_ROWS
sysbench/tests/oltp/sb_oltp.c
29:#define INSERT_ROWS 10000
347:    nrows = INSERT_ROWS;

-- 
You received this question notification because you are a member of
sysbench-developers, which is an answer contact for sysbench.