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: Periodic restart plusses and minuses?

Re: Periodic restart plusses and minuses?

From: Howard J. Rogers <howardjr_at_dizwell.com>
Date: Tue, 05 Oct 2004 15:00:44 +1000
Message-Id: <41637ca7$0$22860$afc38c87@news.optusnet.com.au>


Hans Forbrich wrote:

> I'd been listening with a moderate amount of amusement, but it's now
> getting out of hand ... (9iR2 & W2K)
>
> One of the local DBAs has his heart set on nightly shutdowns to do cold
> backups. Another DBA wants to go for online backups. (Archive logging is
> up and active.)
>
> The debate has gone all over silly territory and back and is now at:
> "Shutting down the database will waste all the automatic tuning benefits
> achieved from the system finding stability in the cache, etc.. It can
> take hours, even months to get the system stable."
>
> Does anyone have pointers to documents, white papers, books, web sites,
> metalink note, etc. that I can drop on their desks to move them off this
> 'shutdown detunes the database' discussion?
>
> Due to politics, I will not provide an opinion or even make a supportable
> statement in either direction. This needs to be an external (to me)
> reference, and preferrably not an unsupported anecdote.
>
> (Once they're over this, I hope the discussion can get back to a pure
> business benefits discussion!)
>
> Thanks in Advance for helping control the local noise pollution!
> /Hans

Good luck with finding that sort of material.

Personally, and I realise this is not what you were after, I would simply ask "*what* automatic tuning benefits actually exist in 9i R2??". Because apart from the PGA, there aren't any that I can think of!

DBMS_SHARED_POOL.KEEP and a suitable after startup trigger will "pre-warm" the library cache.

The buffer cache will be cold, so physical reads will be needed to re-populate it, true enough. But we know that a 100% buffer cache hit ratio means nothing, and so a 0% hit ratio won't mean much either. And there's LOTS of information out there on that particular topic!!

Ultimately, the only thing they can really do to sort this argument out is to run some performance tests after one or two shutdowns, and compare with normal "after several hours, even months" performance.

And the guy advocating cold backups should be asked why he wants to do cold backups in the first place, given that he's in archivelog mode and has the choice... is it just a convenience thing? A lack of skills thing? A "I don't really know RMAN too well" thing? Or what??

Regards
HJR Received on Tue Oct 05 2004 - 00:00:44 CDT

Original text of this message

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