Poor synthetic full performance with StoreOnce?

[Vejete]

I'm seeing some unexpected behavior with Veeam 9.5 and synthetic full backup with HPE StoreOnce.

Working with one (1) VM
The StoreOnce is a 3540 running 3.16.3 code
Active full VM backup to NAS runs 7 hours - that's our basis for comparison
Starting with 1 full and 1 incr, a synth full to StoreOnce runs 12 hours ???
Seeing average speed of 300 MB/s with peaks around 500-600 MB/s during synth full to StoreOnce

The question is: If Veeam runs an incr then sends a plan via Catalyst API to StoreOnce for synth full, why does it take StoreOnce longer to process metadata and assemble new Catalyst media comprised of only pointers to existing data blocks? I guess the perception is that the synth full to StoreOnce should have outperformed the active full to NAS.

Thanks as always!

[foggy]

Hi Jim, basically it should be faster, indeed. However, we've seen a similar case recently, where metadata processing issues led to the same performance decrease. If would be extremely helpful if you could open a case with Veeam support and provide logs for investigation. Thanks!

[jjacobs1]

Hi Foggy, we have been having StoreOnce problems for the past month. Wondering if we have reached a threshold of metadata. It was working fine for months and then stopped. We have an ongoing Veeam case (#02319002).
Seems like the StoreOnce Plugin is timing out.

[foggy]

Hi Joshua, please continue investigating with the help of your support engineer, since it is not clear what sort of issues are you seeing.