Skip to contentSkip to conversation

Small ZFS / Reiser4 / Ext 'benchmark'

DISCUSSION Adrian Ulrich 7 participants 7 messages Feb 2, 2006 – Feb 3, 2006

lore
Share & tools
lore
#0 lore raw

Hi,

If anyone is interested:
I ran a small filesystem benchmark on my x86 PC.

It includes:

On Linux:
* Reiser4
* ReiserFS
* Ext3

On Solaris (Using 'gnusolaris'[.org] -> Alpha 2)
* UFS
* ZFS

NetApp's 'Postmark' was used to perform the tests. (Postmark simulates something like Mail/NNTP-Server load)

Results:
http://spam.workaround.ch/dull/postmark.txt

(I used the *default* mkfs/mount options for all filesystems.
If you like, i can re-run the test with non-default parameters)

-- Adrian

#1 lore raw

Dnia Thu, 02 Feb 2006 22:59:43 +0100, Adrian Ulrich <[email protected]> napisa³:

Quoted 15 lines

If anyone is interested:
I ran a small filesystem benchmark on my x86 PC.

It includes:

On Linux:
* Reiser4
* ReiserFS
* Ext3

On Solaris (Using 'gnusolaris'[.org] -> Alpha 2)
* UFS
* ZFS

Quoted 5 lines

NetApp's 'Postmark' was used to perform the tests. (Postmark simulates something like Mail/NNTP-Server load)

Results:
http://spam.workaround.ch/dull/postmark.txt

Is it just me or did You also found out that reiser4 is "little" faster in those tests?? ;) (Now I know why I'm using r4)

£ukasz Mierzwa

#2 lore raw
In reply to Adrian Ulrich
Quoted 20 lines · Adrian Ulrich

Hi,

If anyone is interested:
I ran a small filesystem benchmark on my x86 PC.

It includes:

On Linux:
* Reiser4
* ReiserFS
* Ext3

On Solaris (Using 'gnusolaris'[.org] -> Alpha 2)
* UFS
* ZFS

NetApp's 'Postmark' was used to perform the tests. (Postmark simulates something like Mail/NNTP-Server load)

Quoted 6 lines

Results:
http://spam.workaround.ch/dull/postmark.txt

(I used the *default* mkfs/mount options for all filesystems. If you like, i can re-run the test with non-default parameters)

WOW! I am misreading something or Reiser4 is _really_ _that_ _fast_?

Could you also add some basic description what these tests do? Do they ensure that the data is really written to disk before the timer stops? What about using bigger partition (and data written/read) - say 4 times your RAM? Are you rebooting after creating fs? Could you make scripts you used available?

Could you add ext2, jfs, xfs to these benchmarks? Maybe also some other bechmarking program? Also maybe add Reiser4 with compression plugin to the list of filesystems.

Also, maybe you should post the updated results (preferably attached, not URL if it will fit in 60KB) to LKML for wider discussion about why things are like this?

Thanks,

Grzegorz Kulewski

#3 lore raw
In reply to Adrian Ulrich

I'm curious if disabling ZFS checksumming would make a significant difference.

Quoted 30 lines · Adrian Ulrich

Hi,

If anyone is interested:
I ran a small filesystem benchmark on my x86 PC.

It includes:

On Linux:
* Reiser4
* ReiserFS
* Ext3

On Solaris (Using 'gnusolaris'[.org] -> Alpha 2)
* UFS
* ZFS

NetApp's 'Postmark' was used to perform the tests. (Postmark simulates something like Mail/NNTP-Server load)

Results:
http://spam.workaround.ch/dull/postmark.txt

(I used the *default* mkfs/mount options for all filesystems.
If you like, i can re-run the test with non-default parameters)

-- Adrian

Signature · 1 line
Jake Maciejewski <[email protected]>
#4 lore raw
In reply to Adrian Ulrich
Quoted 15 lines · Adrian Ulrich

If anyone is interested:
I ran a small filesystem benchmark on my x86 PC.

It includes:

On Linux:
* Reiser4
* ReiserFS
* Ext3

On Solaris (Using 'gnusolaris'[.org] -> Alpha 2)
* UFS
* ZFS

Quoted 8 lines

NetApp's 'Postmark' was used to perform the tests. (Postmark simulates something like Mail/NNTP-Server load)

Results:
http://spam.workaround.ch/dull/postmark.txt

(I used the *default* mkfs/mount options for all filesystems.
If you like, i can re-run the test with non-default parameters)

If you could format (or tune2fs) the ext3 filesystem with "-O dir_index" this would likely improve performance if the test is creating many files in the same dir. Also, is the file size limit in bytes, or kilobytes? Unfortunately, the canonical postmark URLs I can find are not useful.

What is a interesting, though maybe not terribly surprising is that ZFS is doing so poorly in the second test. I'd be extremely interested in seeing the vmstat output while the tests are running, as I've heard that ZFS is CPU hungry because of the checksumming.

Cheers, Andreas
--
Andreas Dilger
Principal Software Engineer
Cluster File Systems, Inc.

#5 lore raw

I am surprised that Reiser4 does so well, and I would be interested in knowing more details of exactly what the test does. Our tests show reiser4 not doing THAT much better, so you must be doing something different from our tests.

I learn from almost every benchmark done by other people, because they almost always test differently from what I had imagined I should test.

Hans

#6 lore raw

Hans,

Quoted 4 lines

I am surprised that Reiser4 does so well, and I would be interested in knowing more details of exactly what the test does. Our tests show reiser4 not doing THAT much better, so you must be doing something different from our tests.

Unfortunately LKML was not CC'd even once in this thread so your mail that CC'd lkml looked very strange because it was a reply to messages not seen on the list :-(

If you had the intention of letting folks outside reiserfs-ml know, you might have missed it. Maybe Grzegorz Kulewski's suggestion about Adrian doing repetetive benchmarks and posting them regularly is quite valid and sensible.

Signature · 2 lines
Best regards,
Pysiak