Re: Poor Backup Strategy???

From: John Claxton <claxtojr_at_email.usps.gov>
Date: 1996/01/12
Message-ID: <4d6138$2pd_at_blue.usps.gov>#1/1

petersen_at_com1.dwhl.de (Helge Petersen) wrote:
>Steve Butler <sbut-is_at_seatimes.com> writes:
>
>
>>On 10 Jan 1996, Glenn Burton wrote:
>>> The system administrators of one of the systems I work with do not shut down
>>> the Oracle instance to take a cold backup. Instead they simply copy the
 

>>If they follow the approved hot backup strategy then everything should be

      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^


>>fine.  This strategy flags the database that indicated tablespaces are 


>>being copied.  The database will then re-synch the thread counters in the 

>>tablespace header when the script indicates the backup is finished (with
>>the ALTER command).
>
>Note that this means you MUST use
>
>ALTER TABLESPACE xxx BEGIN BACKUP;
>host tar .... (or cp or cpio or ...)
>ALTER TABLESPACE xxx END BACKUP;
>
>for EVERY tablespace.
>
>And you MUST run your instance in ARCHIVE LOG modus
>and you MUST backup the archived redologs.
>
>Hope this helps.
>
>***************************************************************************
>Helge Petersen Datenbase- und Systemadministrator
>Draeger Synematic GmbH D-23558 Luebeck, Germany
>Tel. 0451/882-4044 Fax -3031 Email petersen_at_dwhl.de
>***************************************************************************
Don't forget to backup control file(s) at the same time. And backup ALL database files (system,user,rollback,archive logs, and control files) during the backup.
The current scheme WILL NOT WORK (unless the backup is quick enough to finish before Oracle changes the timestamps and/or SCN numbers in the datafiles.

I agree with the previous response. Try to restore.

Get upper mgmt on your side. Backups to production databases are CRITICAL. Poor backup strategy/implementation/testing can close a companies doors. I worked at a 24x7 shop (mfg type) that used hot backups. We did approx $80,000 US revenue per hour per mill. During the five years I worked there, we had two disks go bad, which required restore. We did. They worked. We had tested,restested, refined, reimplemented, and retested our backup routines. IMHO any production shop should do the same. Failure to do so may be criminal. The position of system administrator (my title) as well as Oracle Database Administrator (another hat I wear) carries responsibilities with all the many rights and privileges [ :) ]. Received on Fri Jan 12 1996 - 00:00:00 CET