这几天老外就mysql 5.6和5.5,MariaDB的benchmark讨论的很热烈,参考http://t.cn/zYoXNND一文,而法国朋友身先士卒,在http://t.cn/zYKgDmk中一连做了多个评测。下面大致归纳下
这些文中的观点
1) 在大名顶顶的percona公司的Peter Zaitsev 一文中(http://www.mysqlperformanceblog.com/2013/02/18/is-mysql-5-6-slower-than-mysql-5-5/)
用到的sysbench参数为:
table-size=1000000 --num-threads=1 --oltp-read-only=on --init-rng=on --max-requests=0 --max-time=300 run
然后出来的比较图为
这个是单线程下的结果,可以看到myql 5.5比5.6和即使5.6关闭掉perfomacne schmea功能都要快点。
而当64个线程的时候,结果图为:
可以看到,mysql 5.5居然依然优于5.6一点。文中的评论也值得一看,其中提到了作者用的评测机器硬件情况。
2) 再来看法国人http://dimitrik.free.fr/blog/archives/02-01-2013_02-28-2013.html#162 这篇文章,这篇文章中对之前他做的评测系列做了说明,分别为:
1) http://dimitrik.free.fr/blog/archives/2013/02/mysql-performance-mysql-56-vs-mysql-55-vs-mariadb-55.html
这篇文章的结果是,mysql 5.6优于mysql 5.6和mariadb 55;而且从结果可以看到,机器越牛B,比如到24,32核,则性能越优;
而在http://dimitrik.free.fr/blog/archives/2013/02/mysql-performance-analyzing-benchmarks-part1-mysql-56-and-mariadb-55.html
一文中,则说到了一些要注意的地方,包括:
1)Performance Schema (PFS) 的开销要注意,最好对比时把5.5,5.6的pfs都关闭,其中5.6是默认打开,5.5是关闭的,会有性能的开销;
2) 当用到16,24核的时候(充分利用),mysql 5.6性能比5.5好不少;
3) 使用jemalloc库
(http://wangkaisino.blog.163.com/blog/static/1870444202011431112323846/)
下面是原文摘录:
don't forget to turn Performance Schema off in MySQL 5.6 if you're comparing it with previous MySQL versions (otherwise it'll not be apples-to-apples comparison)..
if you're looking for a full power TPS results on your HW server - share the same CPU cores for both "client" and "server" (or use another server for "client", but be sure your "client" activity (here sysbench) is not blocked by any HW or SW limits other than MySQL server itself)..
check your MySQL Server configuration settings (mine is here, but I'm not pretending to have the most adapted one, and believe me, you will always be able to find a better tuning for your workload by yourself ;-))
adopt jemalloc library from the beginning -- it may save you a lot of headache time! ;-)
这些文中的观点
1) 在大名顶顶的percona公司的Peter Zaitsev 一文中(http://www.mysqlperformanceblog.com/2013/02/18/is-mysql-5-6-slower-than-mysql-5-5/)
用到的sysbench参数为:
table-size=1000000 --num-threads=1 --oltp-read-only=on --init-rng=on --max-requests=0 --max-time=300 run
然后出来的比较图为
![](https://i-blog.csdnimg.cn/blog_migrate/bf696370519d90cd898fe2304d0aca01.jpeg)
这个是单线程下的结果,可以看到myql 5.5比5.6和即使5.6关闭掉perfomacne schmea功能都要快点。
而当64个线程的时候,结果图为:
![](https://i-blog.csdnimg.cn/blog_migrate/68d351726881798b2db740fb8869c9b0.jpeg)
可以看到,mysql 5.5居然依然优于5.6一点。文中的评论也值得一看,其中提到了作者用的评测机器硬件情况。
2) 再来看法国人http://dimitrik.free.fr/blog/archives/02-01-2013_02-28-2013.html#162 这篇文章,这篇文章中对之前他做的评测系列做了说明,分别为:
1) http://dimitrik.free.fr/blog/archives/2013/02/mysql-performance-mysql-56-vs-mysql-55-vs-mariadb-55.html
这篇文章的结果是,mysql 5.6优于mysql 5.6和mariadb 55;而且从结果可以看到,机器越牛B,比如到24,32核,则性能越优;
而在http://dimitrik.free.fr/blog/archives/2013/02/mysql-performance-analyzing-benchmarks-part1-mysql-56-and-mariadb-55.html
一文中,则说到了一些要注意的地方,包括:
1)Performance Schema (PFS) 的开销要注意,最好对比时把5.5,5.6的pfs都关闭,其中5.6是默认打开,5.5是关闭的,会有性能的开销;
2) 当用到16,24核的时候(充分利用),mysql 5.6性能比5.5好不少;
3) 使用jemalloc库
(http://wangkaisino.blog.163.com/blog/static/1870444202011431112323846/)
下面是原文摘录:
don't forget to turn Performance Schema off in MySQL 5.6 if you're comparing it with previous MySQL versions (otherwise it'll not be apples-to-apples comparison)..
if you're looking for a full power TPS results on your HW server - share the same CPU cores for both "client" and "server" (or use another server for "client", but be sure your "client" activity (here sysbench) is not blocked by any HW or SW limits other than MySQL server itself)..
check your MySQL Server configuration settings (mine is here, but I'm not pretending to have the most adapted one, and believe me, you will always be able to find a better tuning for your workload by yourself ;-))
adopt jemalloc library from the beginning -- it may save you a lot of headache time! ;-)