Hi,
On Tue, Jun 06, 2000 at 09:17:33AM -0700, Chris Mason wrote:
>
> The dedicated log device is really important, and I do have bytes reserved
> to make both that and the multiple filesystems to a single journal work.
>
> Once the dedicated log device is coded, I would rather see the shared
> journal disk idea done through LVM than in the journal code. LVM already
> has all the tools to find the hot spots on the disk, allowing the admin to
> move the i/o hogs to different disks. Mostly, I would prefer not to deal
> with the added complexity in the FS code unless people can find a huge
> list of reasons it would be better than using lvm.
It really does need to be done in the journal layer, for one simple
reason --- performance. You need to make sure that the journal
device is streaming to disk sequentially as much as possible, and that
means that, ideally, you want to be able to interleave blocks from
different filesystems into the same journal disk.
A minimal journal management API to allow that should be fairly simple
to achieve, but because of the write ordering requirements, I don't
think we could do this through the existing LVM code.
Cheers,
Stephen
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Wed Jun 07 2000 - 21:00:25 EST