Zbr's days.
March
Sun Mon Tue Wed Thu Fri Sat
           
25
         
2008
Months
Mar
Oct Nov Dec

About :: TODO :: Blog :: RSS :: Old blog :: Projects :: GIT :: Gallery :: Notes

Tue, 25 Mar 2008

Filesystem contest results.

Interested reader can check out results of the ext2/3/4, reiserfs, reiser4, jfs, xfs and btrfs fight for the first prices in dbench, iozone, postmark, maildir performance bench and simple file creation micro-benchmark.

It does not contain maildir benchmark, I will add it tomorrow or later today, xfs has yet not completed and no graphs.

As a conclusion: nothing major changed since previous contest, new btrfs filesystem behaves not that bad in some cases, but quite slow in others... Nothing changed.

Does it mean, that we need something new?

/devel/fs :: Link / Comments (3)

dup wrote at 2008-03-26 11:16:

I've done some tests with a test framework I'm writing (sorry, not available for now, but very soon) and strangely I do not find the same results as you do for XFS. In my tests XFS is the best performer. I used a debian etch with kernel 2.6.18-3 and the filesystems were created and mounted with defaults values...

Ivan wrote at 2008-03-26 11:28:

Xfs format and mount options is not optimal

".....one allocation group for every 4 GB of capacity in your target block device...."

For your test system:

mkfs.xfs -f -d agcount=75 -l size=64m

and mount with options

logbufs=8,nobarrier,noatime,nodiratime,osyncisdsync

BR

Zbr wrote at 2008-03-26 16:15:

Thanks for suggestion, but that does not help much, here is a postmark graph with your options: http://tservice.net.ru/~s0mbre/blog/devel/fs/2008_03_26.html

It is slightly better, but still much worse than anything else except reisers in postmark. And I blame nobarrier option for performance increase, which is unfair, since all other filesystems have barriers turned on, which is the right thing.

Please solve this captcha to be allowed to post (need to reload in a minute): 56 + 92

Comments are closed for this story.