Disks are so cheap nowdays, just buy more of them and do mirroring in
your OS for your disk storage pools. I did, and I sleep better at
night. Used IBM SSA drawers can be had for peanuts.

Roger Deschner      University of Illinois at Chicago     rogerd AT uic DOT edu



On Fri, 7 Feb 2003, Allen Barth wrote:

>[lobbing armor piercing verbage]
>
> Oh ye of narrow vision and holder of golden horseshoe of hardware luck:
>
>1.  Have you ever lost a stg pool disk?  And before you answer "well ya
>just backup again"
>2. SOME DATA IS IMPOSSIBLE TO BACKUP AGAIN!
>3. I REPEAT!
>
>Along with regular clients, I backup data from Sybase and Oracle via
>SQLBACKTRACK.  Some of this data is an incremental.  In this case
>incremental FROM WITHIN the db server.  IE point-in-time data of pages
>that have changed.  Should that data be lost, there is no way to restore
>beyond what was lost unless another FULL or complete backup has been
>created, but it then could be the case that the full is too late a
>point-in-time.
>
>Yup, I already been down that road, and there isn't any good sights to
>see.  Since then, I use raid-5 storage pools with floating hot spares
>whereever I can.   I know I'm still open to hardware failure issues, but
>the likelyhood of taking a hit is greatly reduced.  Performance hit?  Our
>performance measurement guy saw almost not difference in throughput with
>raid-5 versus non raid-5 in the TSM environment.  Basically says to me
>that the ever famed bottleneck isn't visiting dasd land right now.  Also
>keep in mind that NO storage layout/method/etc can protect against
>corrupted data being written.
>
>--
>Al
>
>
>
>
>"Stapleton, Mark" <stapleto AT BERBEE DOT COM>
>Sent by: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
>02/07/03 08:51 AM
>Please respond to "ADSM: Dist Stor Manager"
>
>
>        To:     ADSM-L AT VM.MARIST DOT EDU
>        cc:
>        Subject:        Re: PMR 02528, 082
>
>
>>>From Steven Schraer:
>>>It looks like for storage pools are recommend for raid 0
>>>(mirroring), raid
>>>0+1 (mirroring & stripping) or raid 5 (distributed parity).
>>>These are 0+safe
>>>methods to protect the storage pool.  Do you know of any
>>>companies that use just raid 1 (stripping) on their storage
>>>pools?  Is there an issue of tsm loosing a storage pool and
>>>the database having issues due to the lost storage pool data?
>
>From: William Rosette [mailto:Bill_Rosette AT PAPAJOHNS DOT COM]
>> Does any TSM gurus have any suggestions for our AIX admin
>
>[donning my advocacy armor]
>I still don't see any reason to create redundancy for the disk storage
>pool. Unless you're not using a tape library, there's no reason for it.
>The disk pool should get flushed to a more stable medium, and that flush
>should take place fairly soon after the client backups to disk finish.
>Why waste gobs of disk on something that's going to flushed clear every
>day?
>
>As far the db and log are concerned, just create volume copies with TSM
>and make sure the copies are on disks that are on separate disks.
>
>That's really all there is to it.
>
>--
>Mark Stapleton (stapleton AT berbee DOT com)
>