Ubuntu 12.10 File-Systems: Btrfs, EXT4, XFS

[phoronix]

Phoronix: Ubuntu 12.10 File-Systems: Btrfs, EXT4, XFS

For those curious about the file-system performance of Ubuntu 12.10, here are some benchmarks from Quantal's Linux 3.5 kernel with the EXT4, XFS, and Btrfs file-systems.

http://www.phoronix.com/vr.php?view=17890

[blackout23]

Interesting to the btrfs doing very well with threaded I/O does that make it ideal for SSDs?
Are there stabel fsck tools for btrfs available already?

[russofris]

I've always found it hard to choose a FS based on IO and throughput graphs alone. I wish there were an easy way to incorporate CPU utilization, latency, and other factors without having to resort to fancy OLAP/Cubes or pivot tables. I hate to admit this, but a huge number of our enterprise systems still use EXT3 because I am unable to make a case for another FS in a quick and easily demonstrable manner. Those environments that do run EXT4 were only changed after a duplicate environment was created and full perf/regression analysis had been conducted.

If anyone has an easier way than what is generally accepted as being "the right way", I'm all ears.

F

[TobiSGD]

Not an Ubuntu user, so a small question: Do I have to expect that the Ubuntu kernel will be patched in a way that I can see differences in filesystem performance compared to a stock kernel from kernel.org? If so, wouldn't it make sense to also put the stock kernel into those benchmarks?
Or is there any other reason to test specifically the Ubuntu kernel and not a stock one, besides having Ubuntu in the title of the article for getting more page-hits from Ubuntu users?

[russofris]

Interesting to the btrfs doing very well with threaded I/O does that make it ideal for SSDs?
Are there stabel fsck tools for btrfs available already?

BTRFS has an SSD mount option which changes some of the underlying logic. To date, I have not seen a compelling set of benchmark results to call it faster than without the mount option on an SSD.

There are tools for BTRFS, including fsck. I do not know that I would call them stable (It's a relative term), nor do I believe that fsck is as essential as it was with the previous generation of filesystems.

F

[89c51]

BTRFS has an SSD mount option which changes some of the underlying logic. To date, I have not seen a compelling set of benchmark results to call it faster than without the mount option on an SSD.

There are tools for BTRFS, including fsck. I do not know that I would call them stable (It's a relative term), nor do I believe that fsck is as essential as it was with the previous generation of filesystems.

F

Wasn't the btrfs fsck incompatible with the other ones (ext4 etc) ??

[russofris]

Wasn't the btrfs fsck incompatible with the other ones (ext4 etc) ??

Yes. BTRFS and EXT4 are two different file system with completely different architectures, so this is expected.

F

[russofris]

Not an Ubuntu user, so a small question: Do I have to expect that the Ubuntu kernel will be patched in a way that I can see differences in filesystem performance compared to a stock kernel from kernel.org? If so, wouldn't it make sense to also put the stock kernel into those benchmarks?
Or is there any other reason to test specifically the Ubuntu kernel and not a stock one, besides having Ubuntu in the title of the article for getting more page-hits from Ubuntu users?

I do not believe that there will be performance impacting differences (you could look at the diffs). Regarding performance, I would recommend that you double check the BTRFS mount options during the installation of Ubuntu, as the defaults are pretty tame, and may not be the most suitable for your installation.

F

[89c51]

Yes. BTRFS and EXT4 are two different file system with completely different architectures, so this is expected.

F

Yes of course thats true but i remember reading that it couldn't be used as the other ones hence you cant just fsck as with an ext4 for example.

[ssam]

between the COW and the online checking, there is less need for fsck on btrfs compared to ext3/4. The is a btrfsck since a few months ago:
https://btrfs.wiki.kernel.org/index.php/Btrfsck
and also
https://btrfs.wiki.kernel.org/index.php/Restore
If you do hit file system errors it is worth getting on the btrfs irc channel. they can help figure out if you have actually hit a bug in btrfs.

@russofris
for ext4 you could argue based on features. the main ones being less fragmentation, faster fsck, and more recently metadata checksums https://ext4.wiki.kernel.org/index.p...data_Checksums http://kernelnewbies.org/Linux_3.5#h...a7b417333147f8

for btrfs you could argue that full checksumming makes the data much more safe. with traditional raid1 if a block differs between the 2 drives you only know that something is wrong (and you only notice when scrubbing), with brtfs 'raid1' you can see which version of the block has the right checksum and you can spot corruption without having to scrub.
but the counter argument is that much newer code makes it very unsafe. i am sure some people wont consider btrfs safe until they know it has been in wide spread use for 5 years.