# InnoDB performance and scalability blog

**URL:** <https://forums.percona.com/t/innodb-performance-and-scalability-blog/2871>\
**Category:** Other MySQL® Questions\
**Created:** [August 12, 2013, 4:53am UTC](https://forums.percona.com/t/innodb-performance-and-scalability-blog/2871 "2013-08-12T04:53:39Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![himanshu](https://avatars.discourse-cdn.com/v4/letter/h/ac91a4/32.png) [@himanshu](https://forums.percona.com/u/himanshu)\
**Post date:** [August 12, 2013, 4:53am UTC](https://forums.percona.com/t/innodb-performance-and-scalability-blog/2871/1 "2013-08-12T04:53:39Z")

</div>

When we talk of innodb performance tuning varaibles : innodb\_file\_per\_table, innodinnodb\_buffer\_pool\_size  
innodb\_log\_file\_size, innodb\_flush\_log\_at\_trx\_commit , innodb\_flush\_method…it is seen that setting up the optimum values for these values with respect to your server configurations always fine tune innodb tables…  
But very recently i faced an Issue , I configured these variables in my master DB, which was earlier running fine with default values for about 2 years now….But I did it since the size of Innodb tables have increased drastically and variables were not well tuned… Below given are the values which i configured (My Server status is :: OS – Redhat 5.8 , 16GB ram, 8 CPUs)….

innodb\_file\_per\_table=1  
innodb\_buffer\_pool\_size = 2048M  
innodb\_additional\_mem\_pool\_size = 256M  
innodb\_log\_file\_size = 512M  
innodb\_flush\_log\_at\_trx\_commit=2  
innodb\_flush\_method=O\_DIRECT

But what i saw was, as soon as the hits started pouring in, the connections rose to about 700 and that was persistent until i restarted my Mysql and within mins it rose again….  
Maximum connections was set up at 2000, but we never faced the connections issue until this time. And it was limited to about 150-200 every secs.

AND finally when i commented out the values for innodb\_flush\_log\_at\_trx\_commit and innodb\_flush\_method…  
The performance came back to normal and it was functioning well….Then what could be the reason for this undue behaviour…

thanks…

---

<div class="post-metadata">

**Author:** ![yogesh777](https://avatars.discourse-cdn.com/v4/letter/y/87869e/32.png) [@yogesh777](https://forums.percona.com/u/yogesh777)\
**Post date:** [August 14, 2013, 12:23am UTC](https://forums.percona.com/t/innodb-performance-and-scalability-blog/2871/2 "2013-08-14T00:23:53Z")

</div>

Set innodb\_buffer\_pool\_size to 10 GB if you are not running any other application on this server

> [@](#):
>
> AND finally when i commented out the values for innodb\_flush\_log\_at\_trx\_commit and innodb\_flush\_method… The performance came back to normal and it was functioning well

innodb\_flush\_method affect performance if database have huge read/write operation so you need to configure it according to your storage type i.e. SAN or Local Disk.  
Basically O\_DIRECT open files directly without using caching so reduce double buffering of disk as well as mysql. You need to benchmark your database before setting any flushing method on your storage.