ext3 block size question

ext3 block size question

[jshankar]

Hello,

I was measuring the performance of ext3 file system. As the block size 
increases,read and write performance improves.  why there is a better 
performance improvement for large block sizes??.

Thanks
Jayshankar

Re: ext3 block size question

[Zachary Peterson]

So it's my understanding that by increasing the block size, you guarantee
the contiguity of files.  For instance, GPFS uses a 256K block to achieve
high performance for parallel and cluster computing applications.  It
essentially reduces the number of seeks required to read a large file.
(Seek time being one of the largest components of IO latency.)

There's a drawback, however, in that large blocks incur internal
fragmentation, or the unused storage space contained in a block.
Consider the 1K file being backed by 256K block.  Also, there's poor
integration with the page cache and virtual memory by virtue of a mismatch
in page/block sizes.  Large blocks have to be pined in memory or the
memory system has to perform read-modify-write.

This is why many file systems opt for a block size that's small, usually
the same size or smaller than the page size.  This puts a lot more
importance on allocation techniques to insure block contiguity.


Zachary

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Zachary Peterson       zachary@cse.ucsc.edu
                       http://znjp.com

856D 29FA E1F7 DB5E 9215  C68D 5F0F 3929 C929 9A72

On Mon, 25 Aug 2003, jshankar wrote:

>Hello,
>
>I was measuring the performance of ext3 file system. As the block size
>increases,read and write performance improves.  why there is a better
>performance improvement for large block sizes??.
>
>Thanks
>Jayshankar
>
>-
>To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
>the body of a message to majordomo@vger.kernel.org
>More majordomo info at  http://vger.kernel.org/majordomo-info.html
>