Forever Incremental Mode and GFS to tape

[Vladimir_AUS]

Greetings
I am running Forever incremental Mode do disk, and then doing GFS to tape from disk. Best Practices say that it will create synthetic full periodically on tape. How periodically is it? First synthetic was huge. Therefore my asking.

[Shestakov]

Hi Vladimir!
Virtual full backup schedule is to be set in the job wizard.
You can configure schedule for virtual full backup only if you selected a simple media pool(not GFS one) though.

[Vladimir_AUS]

Hi Nikita,
Let me try again
I am running Forever incremental Mode do DAS, with 7 restore points.
Also, I am running GFS pool and job from DAS to tape with standard retention periods.
Sometimes, weekly GFS is creating synthetic full on tapes, and sometimes not.
My question is - how from a chain of one full and 6 incrementals on DAS weekly GFS to tapes are created. How monthly are created?

[veremin]

GFS backup to tape either creates a virtual full backup (directly on tapes) or takes a full backup from source repository if it corresponds to the required date.

In your case (forever incremental chain) the GFS backup to tape job should always create a virtual full backup. Do you see any messages in the job log, when it does no do so?

Thanks.

[Vladimir_AUS]

Thanks Vladimir,
I had a look and you are correct, all backups are synthetic, except of jobs where full backups were taken, for some reasons.
Would you mind me asking, if I will change Forever Incremental to Incremental with synthetic full to the disk, will it improve performance of GFS job to tape?

[veremin]

I had a look and you are correct, all backups are synthetic, except of jobs where full backups were taken, for some reasons.

What operation was faster? Virtual full backup or full backup archival? Depending on the answer, you will get an idea whether switching mode would result in improved performance.

[Vladimir_AUS]

Thanks Vladimir, archival was faster

[veremin]

Then, switching to archiving existing full backups sounds like a good idea, as that would indeed lead to faster archival speed.