RE: San & single point of failure

  • From: "QuijadaReina, Julio C" <QuijadJC@xxxxxxxxxxxxxxx>
  • To: "pythianbrinsmead@xxxxxxxxx" <pythianbrinsmead@xxxxxxxxx>
  • Date: Thu, 20 Nov 2008 09:59:56 -0500

Valid points! My point was that many storage admins go ahead with configuring 
storage without checking the implications of said configuration on the 
applications that will use that storage.


From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On 
Behalf Of Mark Brinsmead
Sent: Tuesday, November 18, 2008 9:30 PM
To: QuijadaReina, Julio C
Cc: piontekdd@xxxxxxxxx; czeiler@xxxxxxxxxx; oracle_l
Subject: Re: San & single point of failure

What?  Why beware of that?  In some places I have been, this might be the good 
stuff!  :-)   (And yes, I do subscribe to BAARF.)

What you really need to worry about is the situation where some idiot gives you 
a  RAID-1 where the two mirrored volumes are at either end of the same disk!  
Not only is there no protection, but the performance is the stuff of nightmares.

Sadly, many RAID vendors actually claim that RAID-5  (well, their RAID-5, 
anyway) is just as good as RAID-1.  Or better.  And many storage administrators 
will believe them.  Often there is nothing you can do about this.

Happily, I have yet to encounter a storage vendor who claims that it is a good 
idea to do RAID-1 with two slices of the same physical device.  But every now 
and again, you may see a storage administrator do it.  Usually by accident, I 
hope.  Thankfully, I have never encountered this one first hand, but I've seen 
close.  :-)

On Tue, Nov 18, 2008 at 9:10 AM, QuijadaReina, Julio C 
<QuijadJC@xxxxxxxxxxxxxxx> wrote:
Beware also of SAN administrators who setup RAID 5 and later give that storage 
to DBA's for their databases. It does not hurt to double check what kind of 
underlying RAID they've got for you.



-- Mark Brinsmead
  Senior DBA,
  The Pythian Group

Other related posts: