| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Mailing Lists -> Oracle-L -> RE: Rman/Legato/Architecture
Darren/all,
This sort of functionality (spooling to disk) was already available long ago in ADSM (IBM's Tivoli). Not sure if RMAN makes use of it, although I would guess that it does (anyone using RMAN and ADSM?). Btw, ATA disks are relatively cheap, even when stacked up in an array, and Serial ATA (SATA) may even overcome some of the performance issues. Storage industry pundits are even predicting that Tapes as a backup medium may someday fully be replaced by ATA Disks, and I think we are seeing an example of disks replacing tapes in an 'intermediate' storage situation...
Matt Zito may have more to say on this... Matt?
John Kanagaraj <><
DB Soft Inc
Phone: 408-970-7002 (W)
Grace - Getting something we do NOT deserve Mercy - NOT getting something we DO deserve Click on 'http://www.needhim.org' <http://www.needhim.org'/> for Grace and Mercy that is freely available!
-----Original Message-----
From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org]
On Behalf Of Browett, Darren
Sent: Friday, June 11, 2004 9:41 AM
To: oracle-l_at_freelists.org
Subject: RE: Rman/Legato/Architecture
For us staging has performed exceptionally well, our backup window has
shrunk considerably, simply because
backing up to disk is fast. We use a product call Ataboy, produced by a
company called NEXSAN for our disk storage
It has 2.5 Tb of storage. They also have one called the ATABeast, 12Tb.
I installed a second HBA in the backup server, unders windoze, added a new drive, created a directory called backup, then under legato created devices/volumes that use that storage. Created disk pools, pointed groups to use the disk pools as opposed to tape pools. Then create stages, that monitor the disk volumes and send staged data to the tape pool based on the stage policies. Backups work exaclty the same, issue a save to a group and it puts in on disk.
The only problem sometimes is, the when a stage is happening (disk to tape)
it locks the disk volume
so you can't access it. I resolved this by creating smaller disk volumes
(200 GB) as a opposed to larger
volumes (>500 Gb). Also the high water mark doesn't work very well, but
then again I am still on legate 6.1.x. I am
sure some of these issues have been resolved under version 7.
Restores are performed exactly the same, whether it is a disk volume or a
tape volume, it is a volume, simple.
Obviously tape is slower to recover from, and it may not be readily
available.
Darren
-----Original Message-----
From: Fuad Arshad [mailto:fuadar_at_yahoo.com]
Sent: Friday, June 11, 2004 9:18 AM
To: oracle-l_at_freelists.org
Subject: RE: Rman/Legato/Architecture
interesting ,
we are actually thinking about using this. what are your thoughts .
how are restores performed if data is moved to tape.
"Browett, Darren" <dbrowett_at_coquitlam.ca> wrote:
Legato has a feature called staging, which we are using. In short,
we backup up to disk, then based on age of the files, 2 days, 5 days,
7days,
the savesets are moved to tape. Works great.
Darren
-----Original Message-----
From: Stephen.Lee_at_DTAG.Com [mailto:Stephen.Lee_at_DTAG.Com]
Sent: Thursday, June 10, 2004 9:47 AM
To: oracle-l_at_freelists.org
Subject: RE: Rman/Legato/Architecture
I'm not a Legato expert but ...
I think a big part of the problem is that during backup, rman can shovel
the
files into Networker in parallel, but when it is time to restore the
files,
the parallelism (or most of it) goes away. We have started keeping a
copy
of the backup on disk as well as on tape. We have just started this for
the
same reason you have experienced, so we are still experimenting. Our
current strategy -- subject to change if somebody comes along with
something
better -- is to backup to disk; then run a script that uses mminfo to
see
what is on disk that isn't on tape; then use the save command to send
the
new files to tape; then delete yesterday's backup from disk. In our
case,
we have about 1.8 Tb allocated for this. So yeah, we're talking about a
wee
bit of disk space here.
FAQ is at http://www.freelists.org/help/fom-serve/cache/1.htmlput 'unsubscribe' in the subject line.
-----------------------------------------------------------------
----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to: oracle-l-request_at_freelists.org
 |
 |