login
Header Space

 
 

Re: More Large blocksize benchmarks

Score:
Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]
To: Chris Mason <chris.mason@...>
Cc: David Chinner <dgc@...>, Linus Torvalds <torvalds@...>, Nathan Scott <nscott@...>, Andrea Arcangeli <andrea@...>, Nick Piggin <nickpiggin@...>, Christoph Lameter <clameter@...>, Mel Gorman <mel@...>, <linux-fsdevel@...>, <linux-kernel@...>, Christoph Hellwig <hch@...>, William Lee Irwin III <wli@...>, Jens Axboe <jens.axboe@...>, Badari Pulavarty <pbadari@...>, Maxim Levitsky <maximlevitsky@...>, Fengguang Wu <fengguang.wu@...>, swin wang <wangswin@...>, <totty.lu@...>, <hugh@...>, <joern@...>
Date: Monday, October 15, 2007 - 10:36 pm

On Mon, Oct 15, 2007 at 08:22:31PM -0400, Chris Mason wrote:

Apples to oranges, Chris ;)

btrfs linearises writes due to it's COW behaviour and this is trades
off read speed. i.e. we take more seeks to read data so we can keep
the write speed high. By using large blocks, you're reducing the
number of seeks needed to find anything, and hence the read speed
will increase. Write speed will be pretty much unchanged because
btrfs does linear writes no matter the block size.

XFS doesn't linearise writes and optimises it's layout for a large
number of disks and a low number of seeks on reads - the opposite
of btrfs. Hence large block sizes reduce the number of writes XFS
needs to write a given set of data+metadata and hence write speed
increases much more than the read speed (until you get to large tree
traversals).

The basic conclusion is that different filesystems will benefit in
different ways with large block sizes....

Cheers,

Dave.
-- 
Dave Chinner
Principal Engineer
SGI Australian Software Group
-
Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]

Messages in current thread:
More Large blocksize benchmarks, Chris Mason, (Mon Oct 15, 8:22 pm)
Re: More Large blocksize benchmarks, David Chinner, (Mon Oct 15, 10:36 pm)
Re: More Large blocksize benchmarks, Chris Mason, (Tue Oct 16, 9:01 am)
Re: More Large blocksize benchmarks, Christoph Lameter, (Mon Oct 15, 8:44 pm)
speck-geostationary