Filer backup performance

[faridux]

Hello,

I was wondering which is best way to backup several volumes on a filer

for example, let's say there are 20 volumes on the filer and we have 3 drives to back them up

1st option: put the 20 volume paths in one policy, check allow multiple datastream, set the maximum concurrent drives to 3

2nd option:create 20 policies (one for each volume) and set the maximum concurrent drives to 3

I believe the second option avoid the parent job and stream discovery process, and each policy has its own ndmp client session

I do not find any documentation who compares the two options, please could you share your feelings

Thank you

[Nicolai]

Option 1: Verify difficult to spot single share failures because the error does not't show up as a Netbackup status code.

Option 2: As each share has it own policy, it's easy to spot failures. If the share fails the job fails.

Go for option 1 for easy of configuration. Go for option 2 for maximum security regarding errors, but will cost you some additional policy configuration.

[Yasuhisa_Ishika]

When job fails and you have to re-run your backup manually, you can choice which policy(=volume/share) you are to restart with option 2 configuration.

[Zak_Meekins_2]

I vote for option 1 - you shouldnt get that many fails! If you do get a fail you could just run it out of an idel policy.

[Douglas_A · Accepted solution]

Option 1 - This would allow for faster backups overall since you wouldnt have to remount tapes for each backup stream, your job can continue un interupted. downside as stated you would need to keep an eye out for failures and rerun the single volumes as needed.

Option 2 - Alot of policies are hard to maintain, each stream would require remounting tapes causing delays in your backup window.

I would say Option 1 for a more fluid backup and easier managment.

[faridux]

tank you for your advices

I guess the option 1 will be the choosen 1 as it requires less workload,