On Fri, 7 Mar 2003, Eric Sandeen wrote:
> I think it's important to remember that XFS has been in use for -many-
> years on irix,
Yes, we use it there as well. It's just that it came a bit as a surprise
that the choice of a FS would affect the lifetime of the disk...
I guess that we'll soon see disk-savers that save the oxide from our disks
as the screen-savers saved the phosphor from our monitors :-)))
> (And, I can think of other disk usages patterns which could result in
> disproportional use of one area of a disk - /var/log partitions, for
> example).
Swap partitions might be hit even harder...
> I think you're asking too much here...
Well, I should have enclosed the whole block within <nice idea> quotes or
something. But there _are_ performane freaks that pay attention to these
details... Even more the problem might become hard when thinking about
relocated sectors in which case the FS has no idea what is the real
location.
> I think each individual disk would need to be profiled to even know
> where the "fast" spots are.
Oh, I thought that this is common knowledge. Look at
http://www.storagereview.com or
http://www.hardware.fr/articles/448/page1.html (in French)
and you'll see that there are drives which provide at the end of the disk
close to half of reading speed from the beginning of the disk, while
writes are even more affected. So, would you place the journal at the end
of such a disk ? Of course, seek times also count...
> And then technically, the log could go anywhere, but I think that for
> the effort involved, the returns would not be that great.
No, you'r right, possibly the fastest solution is to have a fast SCSI disk
only for the journal and a replacement disk at hand.
--
Bogdan Costescu
IWR - Interdisziplinaeres Zentrum fuer Wissenschaftliches Rechnen
Universitaet Heidelberg, INF 368, D-69120 Heidelberg, GERMANY
Telephone: +49 6221 54 8869, Telefax: +49 6221 54 8868
E-mail: Bogdan.Costescu@xxxxxxxxxxxxxxxxxxxxx
|