[p4] HELP!!!

[Russell C. Jackson]

I have run a tool that I got from support called fstst, and it showed 
the SAN volume that I have my metadata on as being faster than 
comparable DAS storage, so I would not move my db files to local storage 
without testing first. I asked support for measurements on where they 
have proven DAS to be faster than SAN, and they don't have them. They 
are working off of what many customers have told them. That isn't valid 
information. If you have a properly configured SAN attached to your 
system through fiber channel, it can, and often will out perform direct 
attached storage.

If you cut the size of your db files that much, most of the performance 
gain you are going to see is going to be coming from the rebalaced 
B-Trees in your database unless your local disk is tremendously faster 
than your SAN.

Thanks,
Rusty

--------------------------------------------------
RCJackson Consulting
Perforce Consulting Partner and Certified Trainer
--------------------------------------------------
rusty at rcjacksonconsulting.com
http://www.rcjacksonconsulting.com
tel:      479-696-9710
fax:      479-967-0963
mobile:   479-747-3845
--------------------------------------------------




Venters, Cheryl wrote:
> We had originally planned to just move the db files but our performance
> was so bad and our db files so large we thought it would be good to do
> it all at one time. I looked at some of the files that were checked in a
> couple of hours before I made the final checkpoint and they seemed to be
> fine so I think it's okay. I just wanted to double check.
>
> Thanks.
> Cheryl
> -----Original Message-----
> From: Jeff Grills [mailto:jgrills at drivensnow.org] 
> Sent: Saturday, October 21, 2006 11:26 PM
> To: Venters, Cheryl; perforce-user at perforce.com
> Subject: RE: [p4] HELP!!!
>
> You may not have done anything wrong.  Restoring from a checkpoint can
> reduce DB file sizes, and in doing so it can also improve performance of
> your server as well.  I'm surprised that perforce support wouldn't have
> had
> you try rebalancing before moving the files.  And by the way, if you
> weren't
> changing system architectures, you could have just copied your DB files
> instead of restoring from the checkpoint.  The bad news is that you've
> changed two variables at one time, so it's going to be difficult to know
> whether the local files or the DB rebalancing caused any performance
> changes
> you notice.
>
> See http://tinyurl.com/y7r69e or
> http://www.perforce.com/perforce/doc.061/manuals/p4sag/07_perftune.html#
> 1055
> 420 for details about "checkpoints for database tree rebalancing".
>
> j
>
> -----Original Message-----
> From: perforce-user-bounces at perforce.com
> [mailto:perforce-user-bounces at perforce.com] On Behalf Of Venters, Cheryl
> Sent: Sunday, October 22, 2006 1:00 AM
> To: perforce-user at perforce.com
> Subject: [p4] HELP!!!
>
>
> We've been having terrible trouble with performance lately and one of
> the
> things Perforce asked us to do is move our db files to the local drive
> as
> opposed to the san. So, when I did that tonight, I checkpointed my
> original
> server and then restored the checkpoint in the new location. I did a
> comparison of the two directories and noticed that many of the db files
> are
> about half the size of what they were prior to the restore. Is this
> expected
> behavior? One thing that sent me into a minor panic was that I restored
> the
> checkpoint without a corresponding journal file but that should be okay
> since I just checkpointed and there would have been no activity between
> the
> checkpoint and the restore, right?
>
>  
>
> Thanks.
>
> Cheryl
>
> _______________________________________________
> perforce-user mailing list  -  perforce-user at perforce.com
> http://maillist.perforce.com/mailman/listinfo/perforce-user
>
>
>
>