Russ Thompson on 18 Jun 2013 06:51:12 -0700

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: [PLUG] PLUG West - Monday June 17, 2013 - Linux Filesystems and Benchmarking, presented by K.S. Bhaskar (7pm at FIS in Malvern)

For your testing, what IO elevator was used and were any /proc settings tuned? Or is this strictly defaults vs defaults?

Ext4 really shines after some tuning (mainly with SSD's).

- Russ

On Tue, Jun 18, 2013 at 8:59 AM, Rich Freeman <> wrote:
On Tue, Jun 18, 2013 at 7:58 AM, Walt Mankowski <> wrote:
> My understanding is that the main reason ext3 is so much faster with
> mysql (and also with sqlite3) is that barriers are turned off by
> default, while in ext4 they're turned on by default.  Given the choice
> between ext3 (barriers off by default) vs ext4 (barriers turned off by
> choice) isn't ext4 still the better all-around option?

I'm comparing performance on ext3 with barriers to performance on ext4
with barriers.  It may not be the default, but it certainly is
supported (just a mount-time option).

By all means try it both ways for yourself, but everything I can find
online and my own personal experience says that mysql does much better
on ext3 than ext4 when barriers are enabled on both.

That said, if you do want to run mysql on ext4 without barriers then I
would still put it in its own filesystem so that you can still use
barriers with all the data you actually do care about.

Philadelphia Linux Users Group         --
Announcements -
General Discussion  --

Philadelphia Linux Users Group         --
Announcements -
General Discussion  --