Btrfs users - Mar 2008 - btrfs 0.13 and XFS comparison

Hi there,

Just gave 0.13 a spin against XFS (both with default mkfs options) on 
my SATA tower and was impressed by how much performance has improved 
since last time I played with it (v0.5)!

It pretty much matched XFS for I/O performance and had much better 
file creation/deletion performance.

http://www.csamuel.org/2008/03/23/btrfs-013-and-xfs-benchmarks

cheers,
Chris
-- 
 Chris Samuel  :  http://www.csamuel.org/  :  Melbourne, VIC

This email may come with a PGP signature as a file. Do not panic.
For more info see: http://en.wikipedia.org/wiki/OpenPGP
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 481 bytes
Desc: This is a digitally signed message part.
Url :
http://oss.oracle.com/pipermail/btrfs-users/attachments/20080323/62c86882/attachment.bin

On Sunday 23 March 2008, Chris Samuel wrote:
> Hi there,
>
> Just gave 0.13 a spin against XFS (both with default mkfs options) on
> my SATA tower and was impressed by how much performance has improved
> since last time I played with it (v0.5)!
Great to hear, thanks for doing this.
>
> It pretty much matched XFS for I/O performance and had much better
> file creation/deletion performance.
>
> http://www.csamuel.org/2008/03/23/btrfs-013-and-xfs-benchmarks
You can mount btrfs with -o nodatasum to see the performance impact of the 
checksumming code.  It is higher than it should be right now, both reading 
and writing because I'm not interacting as well with the asyncrhonous 
writeback and readahead code in the kernel as well as I should be.

On my 4 drive system here, when I stripe across all 4 drives, I can write at 
180MB/s and read at 200MB/s for the raw device.  Btrfs only does 132MB/s 
writes and 100MB/s reads, but can go at disk speed if I turn off 
checksumming.

(I didn't test XFS, but I fully expect it to go at drive speed).

-chris