Dear list !
I'm reposting this , since got no replies yet.
I need to design a backup policy for a VLDB sized some 10TB, running
as close to 24X7 as possible.
I need 2 versions of the policy:
One is the best case , i.e. money does not matter, the company can aquire
any software / hardware , the
You could split off a mirror and backup the mirror however, I have never
done that so I am not going
to get into it. I know there are others in the list who have done it for
their backup strategy.
Keep in mind that if you are running in archive log mode you need not backup
every data file at
best case, unlimited funds, you need something like EMC's SRDF or IBM's
HACMP which are remote mirrored solutions.
--- Andrey Bronfin [EMAIL PROTECTED] wrote:
Dear list !
I'm reposting this , since got no replies yet.
I need to design a backup policy for a VLDB sized some 10TB, running
thanks a lot Kimberly !
DBAndrey
* 03-9254520
* 053-464562
* mailto:[EMAIL PROTECTED]
-Original Message-
Sent: Wednesday, January 02, 2002 5:35 PM
To: Multiple recipients of list ORACLE-L
You could split off a mirror and backup the mirror however, I have never
done that so I am not
Andrey,
I think that you are looking at this problem from the wrong end. Rather
than come up with backup scenarious first and reporting back to management,
I think you need to get managment to declare to you their Mean Time to
Recovery. By this I mean that they need to tell you how long they
I would also suggest (in a world somewhere between ideal and 'cheapest')
using Veritas NetBackup with the Block-Level Incremental extension for
Oracle. This is really nice for saving on tape costs and backup time - it
performs incremental backups of your datafiles on a block level (thus the
Thanks a lot to all who replied !
Tom , thanks , i forgot to mention , the DB is expected to get recovered in
something like 1-3 hours.
Another correction , the DB size will be around 1TB for the beginning and is
expected to grow to 10TB within a couple of years.
HAPPY NEW YEAR ALL !
The problem is, most business users say 5 minutes (and btw, the emc
srdf option is actually faster than a hot standby, since a hot standby
might need log files to be applied while the mirroring does not --
check out Lawrence To's papers)
so you go ahead and figure out what a 5 minutes scenario
Title: RE: VLDB backup policy
What type of application is it (OLTP, DSS, mixed, etc.)? Also, is it monolithic or can the functionality be broken down into usage profiles (i.e. mostly read, high velocity update, etc.)?
-Original Message-
From: Andrey Bronfin [mailto:[EMAIL PROTECTED
One problem with SRDF though is that it doesn't protect against block
corruption. SRDF is really really cool, but is not a complete backup plan
in and of itself.
George
- Original Message -
To: Multiple recipients of list ORACLE-L [EMAIL PROTECTED]
Sent: Wednesday, January 02, 2002
absolutely true -- but if they want instant recovery, then this will
handle hardware failure
--- George Schlossnagle [EMAIL PROTECTED] wrote:
One problem with SRDF though is that it doesn't protect against block
corruption. SRDF is really really cool, but is not a complete backup
plan
in
It works really well for a DR site though!!
Greg
-Original Message-
Sent: Wednesday, January 02, 2002 2:20 PM
To: Multiple recipients of list ORACLE-L
One problem with SRDF though is that it doesn't protect against block
corruption. SRDF is really really cool, but is not a complete
Title: RE: VLDB backup policy
check out SharePlex for Oracle. The target system is completely open and available while it is applying the changes from the source. If you are expecting 25 GB of redo generation per day, it will usually be under 15 seconds latency. And it's a LOT cheaper than
13 matches
Mail list logo
