Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Oracle on ram drive with no redo log or other archiving

Re: Oracle on ram drive with no redo log or other archiving

From: Connor McDonald <connor_mcdon..._at_yahoo.com>
Date: Tue, 04 Oct 2005 19:05:02 +0800
Message-ID: <434261DE.70DE@yahoo.com>


robertbrown1971_at_yahoo.com wrote:
>
> HansF wrote:
>
> HansF,
>
> Thank you for your answers. I'm putting them to use. I do realize that
> Oracle has very sophisticated caching and thus a lot of operations are
> as fast as if they were in memory especially if the READ/WRITE ratio of
> the application is high.
>
> In my case, though, there is almost as many updates as selects and it
> takes a lot longer to durably store data on harddisk compared to
> storing it on a ram drive. Tests run about 20-25 times faster when
> Oracle storage is in ram already and I'm just looking to get that to go
> even faster by minimizing unneccessary I/O.
>
> - robert
>
> > Why bother with RAM drive under a traditional Oracle database when Oracle
> > already provides a full-fledged optimized (and supported) in-memory
> > database. (Unless you are trying to buffalo some sucker?)
>
> > On Wed, 28 Sep 2005 17:32:42 -0700, robertbrown1971 interested us by
> > writing:
> >
> > > Question:
> >
> > Yee-haw, ride 'em cowboy. Can anyone say 'skunk works'?
> >
> > >
> > > 1) How can I start oracle to minimize or disable redo log? I know there
> > > is a NOARCHIVE mode. Is that it? How do I enable it for the whole
> > > instance?
> >
> > Nope. NOARCHIVE is related to how the log files are safely tucked away by
> > the archiver.
> >
> > Basically the redo log is so fundamental to Oracle's philosophy of
> > protecting the database that I doubt there is a way of telling it to not
> > generate any redo. There are ways of minimizing it, yes, but not
> > eliminating. Then again, is anything wrong with having 2 redo log
> > groups at, say 1M (or less). They are used in a circular buffer basis,
> > wasted pretty quickly, and eliminate any risk of hitting some undocumented
> > log file check/read that might exist (although logically probably doesn't).
> >
> > Perhaps a few days in the Concepts manual will help.
> >
> > > 2) What other logs are there and how can I turn them off (or minimize
> > > output)
> >
> > look for the background_dump_dest (and other *dump_dest) parameters. The
> > alert.log comes to mind. And you may want to point those directories to
> > oblivion.
> >
> > >
> > > 3) What other files besides that data files for the tablespaces used by
> > > tests need to be in RAM drive for fastest performance. In other words,
> > > if I have a regular oracle account and just execute vanilla updates
> > > where does Oracle write to besides the redo log and the data files.
> >
> > Control file, orapwd, spfile/pfile are required to run, or at least start.
> > First one is a writeable file and MUST be visible. This would also be
> > identified in the Concepts manual. (Ya, I know - that takes time, and
> > things are moving real fast here!)
> >
> > >
> > > Note that not all are in RAM. Since ram is expensive it would be
> > > preferrable to keep the files that are changing rarely in durable
> > > location and then only put the files that actually affect performance on
> > > the ram drive.
> >
> > Check out Oracle's Times Ten as well.
> >
>
> > --
> > Hans Forbrich
> > Canada-wide Oracle training and consulting
> > mailto: Fuzzy.GreyBeard_at_gmail.com
> > *** I no longer assist with top-posted newsgroup queries ***

there's plenty of tweaks to make oracle bypass some of its important things, but the question has to be - what are you hoping to prove ? Armed with a ram-based, minimal "io" oracle system, what then?   

-- 
Connor McDonald
Co-author: "Mastering Oracle PL/SQL - Practical Solutions"
Co-author: "Oracle Insight - Tales of the OakTable"

web: http://www.oracledba.co.uk
web: http://www.oaktable.net
email: connor_mcdonald_at_yahoo.com


"Semper in excremento, sole profundum qui variat."

------------------------------------------------------------
Received on Tue Oct 04 2005 - 06:05:02 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US