Re: Datawarehouse backup

> All:
>
> We are planning to cold backup for 1 TeraBytes of datawarehouse
> database on weekly basis (every Saturday).
> Version :9i R2,
> OS: Sun Solaris
> No Archivelog Mode

Why are you planning to operate in NOARCHIVELOG mode? Nobody should *plan* to operate that way.

The only possible reason for operating production in NOARCHIVELOG mode is a proven inability to backup archivelogs due to incredible volume. Has that been proven already, factually?

>
> These are the steps we decide to perform.
>
> A)Backup
>
> 1) Stop ETL jobs
> 2) Change all the tablespaces to READ-ONLY mode except SYSTEM
> 3) Alter system switch logfile( 3 times)
> 4) Shutdown database
> 5) Copy system tablespace datafiles, the control files and the online redo
> logs.
> 6) Startup database
> 7) Compress & Copy all the READ-ONLY tablespaces datafiles to tape
> 8) Change all the READ-ONLY tablespaces to READ-WRITE
> 9) Start ETL job
>
> B) Recovery:
> 1) Backup previous cold backup
> 2) Run ETL jobs up to date

Reliance on ETL jobs for "recovery" is fraught with risk and entails detailed planning, testing, documentation, and flawless execution. Database recovery in ARCHIVELOG mode, on the other hand, is automatic, standardized, supported, and 100% accurate.

>
> The only db outage would be between step # 3 and 5. These steps will
> be controlled from control-M product.
>
> I would like to know, is there any problem in this approach or is
> there any better way to perform the same?

Yes, of course there are problems with this approach and there is a better way: don't run in NOARCHIVELOG mode.

--
http://www.freelists.org/webpage/oracle-l