> BTW, ext3 keeps a kdev_t on-disk for external journals. The
> external journal support is experimental, added to allow people
> to evaluate the usefulness of external journalling. If we
> decide to retain the capability we'll be moving it to a UUID
> or mount-based scheme. So if the kdev_t is being a problem,
> I think we can just break it.
If experience on JFS is any predictor, the external journal will be
quite useful as a performance issue. It can be put on a faster device to
avoid bottlenecks. With JFS I finally wound up with the journal on a
battery-backed SCSI solid state disk, and got about 30% faster completion
of my daily audit run which deleted ~1000000 (yes one million) files as
fast as it could when done.
I was also creating the same number of files over the course of a day,
which is still a respectable directory change rate!
I predict that applications which create/delete a lot of files will run
better with tuning this feature.
-- bill davidsen <davidsen@tmr.com> CTO, TMR Associates, Inc Doing interesting things with little computers since 1979.- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/