[ZODB-Dev] Writeperformance for different storages

[sean.upton@uniontrib.com]

What about putting the journal directory in DirectoryStorage on a
solid-state disk or ramdisk?  Might this help in certain situations with
write performance, considering batching is possible?

Sean

-----Original Message-----
From: Toby Dickenson [mailto:tdickenson@geminidataloggers.com]
Sent: Wednesday, December 18, 2002 6:14 AM
To: Andreas Jung; Barry A. Warsaw
Cc: zodb-dev@zope.org
Subject: Re: [ZODB-Dev] Writeperformance for different storages


On Wednesday 18 December 2002 1:38 pm, Andreas Jung wrote:
> --On Mittwoch, 18. Dezember 2002 08:25 -0500 "Barry A. Warsaw"
>
> <barry@zope.com> wrote:
> > That's not bad actually, considering all the overhead (and benefits)
> > of BerkeleyDB, but that does require gobs of memory and at least two
> > independent disk controllers

I know the BDB documentation says it is a good thing, but did you measure 
whether the independant controllers actually made a difference? to reads or 
writes? I would not expect that ZODB (particularly under ZEO) has enough 
concurrency for this to make a difference.

(If it makes a difference for BDB, I might consider resurrecting a plan to 
stripe a DirectoryStorage across directories on multiple controllers)

> >  (one for the BDB log files and one for
> > the data).  Plus some system-specific tuning of cachesizes, etc.
>
> So your conclusion is: there is nothing faster than FileStorage? :-)

Its hard to imagine a storage writing any fewer bytes to disk, or writing
them 
in a way that is any faster to fsync. I think FileStorage has unbeatable 
write performance.

Semi-serious suggestion: DemoStorage ?


-- 
Toby Dickenson
http://www.geminidataloggers.com/people/tdickenson

_______________________________________________
For more information about ZODB, see the ZODB Wiki:
http://www.zope.org/Wikis/ZODB/

ZODB-Dev mailing list  -  ZODB-Dev@zope.org
http://lists.zope.org/mailman/listinfo/zodb-dev