Re: Used RAID on SQL NT?
Date: 14 May 1994 16:40:02 -0400
Message-ID: <2r3cv2$nql_at_search01.news.aol.com>
In article <CprKKM.Gu0_at_world.std.com>, RAID7_at_world.std.com (John OBrien) writes:
>Subject: Re: Used RAID on SQL NT?
>From: RAID7_at_world.std.com (John OBrien)
>Date: Fri, 13 May 1994 23:14:45 GMT
>Message-ID: <CprKKM.Gu0_at_world.std.com>
>>>used RAID on Windows NT SQL Server and got a performance gain?..
>The new RAID 7 architecture provides high speed I/O access to large
>databases.
*------------------------------------------------------------------------------
John, how does RAID7 intelligently address the classical DBA problem where we optimize our databases for performance based upon an expert (and experientially derived) knowlege of just how our data are being used by the various people and things that call upon our databases?
I can see how some of the simpler issues are addressed, but when it comes to
things like knowing that objects a,b,c in tablespace alpha (made up of
datafiles 1,2,3)
are always accessed and processed whenever d,e,f are accessed -- and therefore,
it would be smart to separate those two sets of objects as widely apart from
each other
as possible on the system-- this is the sort of thing I don't grasp.
In my discussions with our hardware vendor and with our support organization
that runs
our machines, I've discovered it's difficult for me to get this type of issue
across to them.
I'd more than willingly give up control of what Oracle data is put where (which
seems to be
the first thing one must be willing to sacrifice to go to RAID (except for full
mirroring)), if
I could gain an understanding of how the RAID subsystem would perform
optimization
at this level.
Any information you would be able to provide would be most gratefully received.
R. (Scott) Hunter internet:shunter_at_cfotax.attmail.com Database Administrator
AT&T Tax Systems
Received on Sat May 14 1994 - 22:40:02 CEST