Oracle FAQ Your Portal to the Oracle Knowledge Grid

Home -> Community -> Usenet -> c.d.o.server -> Re: backup - openVMS 7.3-1 and Oracle 8.1.7

Re: backup - openVMS 7.3-1 and Oracle 8.1.7

From: Michael Austin <>
Date: Wed, 21 Jul 2004 13:39:10 GMT
Message-ID: <2muLc.16522$>

Syltrem wrote:
> Hi
> My problem with RMAN is cost.
> It needs a full blown tape library to run, AFAIK (unless you're willing to
> first copy to disk using RMAN, then manually backup to tape).

Not true!! As I said we did RMAN disk-to-disk and then had another cluster member move it to 2 copies of the tape (one for onsite and one for offsite storage.) The tape is the last ditch effort should something happen to the 2 full copies plus incrementals.

The details of the backup strategy were actually a LOT more complicated. We used the HP Storage Works EVA technology to create snapshots (logical pointer in the storage system) that pointed to the actual data. It was from this snapshot that the backups were moved to tape. When you have over 200 databases with the smallest in the 3-400GB range, your "cost" perspective is dramatically different. :) This particular data center is approaching ~350TB of SAN storage.

   95% OpenVMS, 5% AIX.

> Pls correct me if I'm wrong, because if I could, I sure would use RMAN.
> At this time I backup to my tape drive. DB size (prod) is 100GB so I can
> still handle it. I do full backups every day.

If you haven't had to do a full restore, you should try it to a test system and see if the cost of downtime can be tolerated for the duration of reading it back from tape. The way they were able to justify the configuration was by handing the technical staff a stack of tapes and told "system is completely toast - here are the tapes - restore the environment." After 30 hrs the test was aborted. They then did disk-to-disk backups, ran the same test and had the environment completely restored and operational in 6-8 hours. With a hourly cost of approximately $250K-300K/hr you can justify an additional $850K for a fully loaded EVA5000 (240 * 146GB drives)and still save money.

Now, how important do you think a fast recovery would be to those shops whose downtime cost in excess of $1MM/hr. That is why those guys use redundant data centers using active-active clusters and have everything mirrored.

Michael Austin.
Consultant - Available.
Donations welcomed.
Received on Wed Jul 21 2004 - 08:39:10 CDT

Original text of this message