We are in the process of re-organizing our current SAN (HSG80), with that I
will be moving
our oracle databases from 3 36Gb drives to 18 9Gb drives. A couple of
scenarios
I have thought about are :
(I should note that these drives will be part of a cluster service which
runs oracle and 9 separate instances.)
1. Within the SAN create an 18 disk raid system.
2. Create 3 separate 6 disk raids and mount them to 3 seperate mount points,
either to
the root directory (/oradb1 /oradb2 /oradb3) or a series of
sub-directories ( /oradb1/app
/oradb1/oradat1 /oradb1/oradat2)
3. Create 3 separate 6 disk raids, combine them using ADVFS and mount to a
single point.
The questions/ comments I have are :
What would be the most effiecient scenario to use: ?
With scenario 1, I have been told it is less efficient that 3 separate
raids, as it could cause contention
across the controller, plus the if one drive fails, it could lead to larger
problems.
With Scenario 3, would the data be spread across all 18 drives or as more
space is taken up the more drives
will be used. ?
With scenerio 2, I would have certain databases on certain volumes based on
there usage,
thus (hopefully) spreading the i/o load around.
I am fairly new to Sys Admin, plus the only Unix person on staff, so please
bear with me if these questions/comments don't
quite make sense, I am learning, especially from this list.
Thanks
Darren ( I will summarize)
----------------------------------------------------------------------------
-------------------------------------------------
Darren Browett P.Eng This message was
transmitted
Systems Admin/DBA using 100% recycled
electrons
Information and Communications Technology.
City of Coquitlam
P:(604) 927 - 3614
E:dbrowett_at_city.coquitlam.bc.ca
----------------------------------------------------------------------------
-------------------------------------------------
Received on Thu Jul 26 2001 - 18:48:33 NZST