Code review comment for lp:~zephyrleaves/percona-server/row_cache

Kevin.Huang (zephyrleaves) wrote :

> I can't get good performance numbers with row cache.
>
> I am running sysbench oltp, ZipF distribution, 100GB database, 16 threads
>
> mysqld --innodb-buffer-pool-size=10GB --innodb-log-file-size=4G
> --innodb_flush_log_at_trx_commit=1 --innodb_row_cache_mem_pool_size=10GB
> --innodb_row_cache_on=ON
>
> with innodb_row_cache_mem_pool_size=10GB the results is actually worse then
> with innodb_row_cache_mem_pool_size=5GB
>
> and in any case it is worse then with no row cache at all.
>
> I suspect it maybe a some mutex contention issue.

You need use this sysbench
http://code.google.com/p/row-cache-for-innodb/downloads/detail?name=sysbench-0.4.8.tar.gz
for test.
This sysbench which i modified for row cache's scene.
You can use command like this to run test:
sysbench --test=oltp --oltp-test-mode=simple --oltp-skip-trx=on --oltp-table-size=80000000 --oltp-range-size=1 --mysql-host=localhost --mysql-user=xx --mysql-password=xx --oltp-read-only=on --init-rng=on --num-threads=70 --oltp-dist-type=special --oltp-dist-pct=1 --oltp-dist-res=80 --max-requests=0 --max-time=1800 run

Because of the common sysbench's oltp-dist-type is not random enough.

« Back to merge proposal