[OpenAFS] improving cache partition performance

Dale Pontius pontius@btv.ibm.com
Mon, 29 Aug 2011 12:08:48 -0400

On 08/29/2011 10:47 AM, Simon Wilkinson wrote:
> To paraphrase a famous AFS saying: The fastest disk access is the one you don't do.
> If your disks are having performance issues (I'd also be interested in seeing utilisation figures, and the difference between the await and svctm times), then the easiest way to fix that is to have less requests hit the disks.
> There are a few ways you can do this
> a) Turn off journalling - you don't need a journalled file system for an AFS cache, and ext3 is known to be slower than ext2
> b) Use a 1.6.0 AFS client. The Linux VM layer in the 1.6.0 codebase has been rewritten, so that we make much better use of the page cache, and satisfy more results from memory rather than going to the underlying disk
> c) Add more memory to your servers. Linux will use any free memory to cache disk data, so this can significantly reduce your IO overhead.
It's possible to turn journaling off in ext4.  Ext4 has much more 
fundamental improvements over ext3, where ext3 basically added 
journaling to ext2.  Does anyone have a feel for running the afs cache 
this way?  This commit seems to indicate that it's a good idea: 

Dale Pontius

Dale Pontius
Senior Engineer
IBM Corporation
Phone: (802) 769-6850
Tie-Line: 446-6850
email: pontius@us.ibm.com

This e-mail and its attachments, if any, may contain confidential and privileged material for the sole use of the intended recipient. Any review, use, distribution or disclosure by others is strictly prohibited. If you are not the intended recipient (or authorized to receive for the recipient), please contact the sender by reply e-mail and delete all copies of this message from your system without copying it and notify sender of the misdirection by reply e-mail.