Duplicate Jobs

[dbaish]

Here's the situation:  I backup a remote server with a backup to disk job.  I then backup that .bkf file to tape.  I read about duplicate jobs and how they're much better than just copying the .bkf to tape.

I'm just wondering if the duplicate job pulls the data from the .bkf file or if the duplicate job pulls the data from the remote server.  The reason I ask is because the backup to disk pull from the remote server is very slow.  However, the backup of the .bkf to tape is pretty fast.  Would the duplicate job be fast because it's pulling from the .bkf or slow because it still pulls from the remote server?

[Untalented]

The duplicate backup jobs pulls from another backup (ie. your bkf) file. So it should be fast.

[mcornes]

Just to give you an idea of the numbers we work with. We backup rougly 1.5TB weekly, that takes about 40 hours to pull from the network to our backup disks and then about 10 hours to backup that to tape using a duplicate job. We're currently experimenting with a second duplicate of that weekly job to a USB disk. I'll report back with timings on that.

[dbaish]

Very informative, mcornes.  Keep us updated with your findings.  Thanks.

[mcornes]

OK Here is the full run down.
The two duplicates took 22 hours running concurrently. The USB connected disk had a job rate of ~1000MB/min compared to the duplication to tape via scsi with a job rate of ~2000MB/min.

To compate the single duplicate of a full job to tape usually takes ~9 hours.

Whilst the jobs were running I successful was able to restore a file the job appeared to hang backupexec but I took my eyes of the screen of a few moments and when I looked back the screen had un hung and the restore had finished. Amusingly the restore happened from the mybook backup. I have since "un-enabled" that device in the all devices pool and a restore paused. So I might be having to somehow remove that device from the place to go for restores. It might be as simple as pausing the device and unplugging it - but  I'll have to look into that / hope someone on here can clue me  in.

give restores could still happen during this time I'm not too fussed, as long as it's finished for tonights job. The next idea is to figure out how to rotate two of those disks and keep them offsite for a month. This will probably come to replace any monthly job that tried to run and always fails because the schedule is pretty jab packed with the weekly full and a daily differential.

m

ps if anyone can confirm how I should remove the first mybook to stop restore jobs try to use it that'd be great.

[mcornes]

I deleted the Backup-to-disk folder from the all devices device pool. The all devices pool must be the default for all our restores and backups.

Removing it from here meant the next test restore went to the normal disks for the file.