Re: anyone have exp w/raw vs. jfs on IBMRS6000 SP2

Not so sure about this. What is said is that Veritas is better than standard file systems. Not that it is better than raw.

Nowadays with modern file systems it is possible to obtain similar results on Read performance between raw and fs.

But Write is always faster in raw than via a fs (abstracting from fs "tricks" to simulate fast writing).

One of the problems with all these evaluations is that no clear distinction is made between Read and Write operations. I've also seen plenty of Write evaluations where the redo logs were kept in fs and tablespaces in raw. This is EXACTLY the opposite of what is needed to make raw perform at its best.

As for the supposed "difficulty" in managing raw partitions, that used to be true
many years ago. Any decent modern UNIX has plenty of excellent tools that make setting up and managing raw easy as pie.

What do I use? File systems for tablespaces, raw for redo logs. Magic performance!

--
Cheers
Nuno Souto
nsouto_at_nsw.bigpond.net.au.nospam
Is there a nospam domain?
http://www.users.bigpond.net.au/the_Den Christopher M. Day <christopher.day_at_rdbms.freeserve.co.uk> wrote in message news:37715D4A.43F59049_at_rdbms.freeserve.co.uk...
> Dan,
>
> Not surprised, seems to consistent across jfs type filesystems ie
> Veritas - might be worth checking http://www.jlcomp.demon.co.uk for his
> comparison article.
>
> Chris
>
> "Daniel P. Considine" wrote:
> >
> > Oracle doc touts raw for performance vs jfs
> > and it is also required for OPS with 7.3.4 and earlier
> > but i am dubious since our jfs dbs are getting better i/o truput than
> > our raw ones
> >
> > assume same database str file system not withstanding
> >
> > thanks for your time
> >
> > Dan Considine
> > Global Information Systems
Received on Wed Jun 23 1999 - 23:20:26 CDT