SOBR full backup placement logic

[ian0x0r]

Hi,

Is there any logic in SOBR that determines if an extent is likely to not be able to accommodate a full backup so will create it on another extent if the SOBR is in data locality mode? If a job is set to synthetic full, there should be a good chance of being able to work out how large the full backup file will be.

The reason I am asking is I've seen an extent fill up, fail the job, the job retry but then write the data to a different extent. If it wrote to the next extent in the first instance then it could have avoided the failure.

I appreciate there are other factors at play here like multiple jobs writing to the same repository etc so maybe not so easy to calculate the space required ahead of the backup job running.

Cheers,

Ian

[dellock6]

If it's not the first initial full backup, one of the criteria used by SOBR is to keep the chain in a single location to limit the failure domain to one single extent, thus the usage of the same extent where the other points of the chain are stored. The job cannot fail before, well, failing... On the retry, the failure marks the extent as full so the retry can use another extent. Indeed it's not easy at all to estimate the final size of an incremental without running the incremental beforehand.