Journaling File Systems and RAID (was RE: [ale] ext3)

[Vernard Martin]

On Wed, 2001-09-19 at 10:44, Michael Perigard wrote:
> 
> Youre question is a little apples vs oranges.
> 
> We run AIX at the office with jfs. We have nfs mounted user directories
> and model directories (housing the engineer's and developer's CADD/CAM
> files) at sizes ive never seen before or even imagined.  The real
> advantage I have seen that jfs has over say, ext2, is the ability to
> extend the filesystem at will, without previously allocating that space to
> a particular volume. By placing several logical volumes in a volume group,
> youre guarenteed that you don't allocate a logical volume more space than
> is needed, allowing you very efficient use of your disk space, RAID array
> or not.
> 
> Allocating your available space to volume groups and then creating and
> managing logical volumes within that group eliminates (almost ;) a full
> filesystem. We have a very simple cron job that email/pages the proper
> people when a filesystem is filling up. This not only keeps the users
> happy and things running smoothly, but alerts us to trouble when we see a
> filesystem run to capacity a few times during a day (a run away log file,
> for example.)
> 
> -Michael

the Logical Volume Management functionality that you describe aboe is
not a feature of journalising filesystems per se but its own separate
capability. You can add LVM capabality to ext with the LVM project
without using a journaling filesystem at all. 

Journaling filesystems primary and only concern is to minimize the
actual time it takes to bring a filesystem up after an unclean shutdown
by storing metadata in such a way that you don't have to do a fsck. This
is not foolproof as you can still loose data when a journaling
filesystem is shut down improperly but the odds are dramatically
improved that you won't.

And regardless of implementation, the amount of data that is written
with a journaling filesystem versus a non-journaling filesystem (keeping
everything else the same) is going to be larger so its going to
technically take longer. The reason you don't notice it that much is
that typically journaling filesystems go through a lot of effort to
employ techniques to speed up writes that normal filesystems don't do
because they already have acceptable performance. 

just my two cents worth

V
--
To unsubscribe: mail majordomo at ale.org with "unsubscribe ale" in message body.