Does first backup to Perf-based SOBR spread across repos?

[pkelly_sts]

If you create a new job, pointing to a new SOBR configured for performance (rather than locality) containing 2 equally sized extents, when the job first runs, will it spread the VMs across the two or will it start using one first, then only switch to the other after a given % usage?

[foggy]

It will place VMs so as to equally utilize the space on both extents.

[pkelly_sts]

Great, I thought/hoped as much, thanks.

[dellock6]

This obviously if both the extents are both configured to accept full backups and incremental backups ( which is a bit against the design of the performance policy, but it works anyway as you described).

[pkelly_sts]

Yes, I did think about that before trying it but my current testing is to make best use of available storage capacity, with a little bonus of having great throughput as well.

What will be interesting to see is the transform performance though as, thinking about it now, I'll still be in a situation where BOTH repositories are hit with simultaneous I/O which I guess is where the concept of performance would make the biggest difference.

However, I did do some testing to a SOBR with one repo for full and the other for incremental and didn't notice any real perforamnce difference (for transforms) compared to just sending everything to a single repo. However, that was with a single big backup file configured whereas for this test I've enabled per-vm backups.

Would you expect much difference in performance to a single full/incr pair of repos if switched to per-vm?

[foggy]

Depends on the storage, whether it benefits from multiple writing streams or not. Some effect may also come from less fragmented backup files.

[pkelly_sts]

Understood. I'll set it to a short enough retention to force it to do a transform over the weekend & see how the performance is.