Hi
Using incrementals on Veritas Netbackup has a disadvantage; it slow down
restores. There is support cases at Veritas confirming that when using
incremental backups, restore times can take up to 10 times normal time.
The solution for this, according to Veritas, is to use full backups
instead of incremental.
And, when using multiplexing, the restore won't go that much faster.
Netbackup still has to do a partial restore;first the full backup, then
the incrementals. This is not a effective way of doing restores...
Best Regards
Daniel Sparrman
-----------------------------------
Daniel Sparrman
Daniel Sparrman
Exist i Stockholm AB
Propellervägen 6B
183 62 HÄGERNÄS
Växel: 08 - 754 98 00
Mobil: 070 - 399 27 51
"Seay, Paul" <seay_pd AT NAPTHEON DOT COM>
Sent by: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
2002-04-18 05:28
Please respond to "ADSM: Dist Stor Manager"
To: ADSM-L AT VM.MARIST DOT EDU
cc:
Subject: Re: Veritas/Legato/ArcServ
I can speak to Veritas NetBackup.
You are correct that a tape is created for each full and incremental, but
there is a feature in NetBackup that allows you to multiplex the server
saves to a single tape up to a limit you specify. This really great until
you try to duplicate the tapes for offsite storage. The duplications tape
literally exponentially longer as the number of files gets larger. We
found
it quicker to run the full backup twice than to duplicate the tapes. The
backup involved ran for about 20 hours on a full to get the two copies.
This was using Magstar tape and Shark disk, big file system on a high end
UNIX SGI machine.
The real issue is NetBackup supports the original and a duplicate only. It
supports 9 retention periods only. It does not have a deleted file
policy,
so once the tapes expire the data is gone. It is tape expiration oriented
instead of object. It has a tape catalog and what files are on the tapes.
When you do an incremental restore there is no directory structure
rebuild,
meaning that files that have been deleted are restored, so the concept of
put the system at point in time does not exist. This can play havoc with
applications using touch files to signal application progressing.
You talk about more tapes, but more tapes also means more tape drives to
process. There is no disk pool concept in NetBackup. It is written
directly to tape, period. This is where TSM eats NetBackups Breakfast,
Lunch, Dinner and Midnight Snack. We took a 12 Magstar drive NetBackup
windows environment using about 180K of NT hardware and migrated it to TSM
using 6 drives and a AIX P660. Nothing else changed except it freed up 6
high end NT servers. Instead of running a few of our 108 servers each
night
on full backups and round robin throughout the week, no C: drive backups,
no
point in time restore capability, duplicates often late getting offsite,
only 2 weeks of recover, we now have a primary, onsite copy, offsite copy
and sit around joking about the days of the NetBackup debacle we
extracated.
The company is about to adopt a daily offsite strategy. Thank you TSM for
making it a possibility.
Plain and Simple. NetBackup is extremely expensive to run. The
maintenance
on the software is much higher than TSM, it costs more than TSM, it uses
at
least twice as much hardware as TSM, has horrible support, and just plain
does not scale.
|
