we have upgraded our backup landscape Enterprise Plus from v10 to v11. We are backing up to a SOBR consisting of 4 extents (Windows 2019, ReFS) with the following each server:
- 512 GB RAM
- 2x AMD EPYC 7252 8-Core Processor 3.10GHz, in total 16 Cores
We write the backups with backup copy jobs to another SOBR, also consisting of 4 extents (Windows 2019, ReFS) with the following equipment:
1x Server with the following:
- 2x AMD EPYC 7252 8-Core Processor 3.10GHz, in total 16 Cores
- 256GB RAM
3x Server with the following each server:
- 2x AMD EPYC 7282 16-Core Processor 2.80GHz, in total 32 Cores
- Fibre Channel connection
- Additional proxy for storage snapshots
The repositories are all set with a "Limit maximum concurrent tasks to: 14", i.e. the SOBR for Backups and Backup Copies then has 4x14= 56 tasks available to each SOBR.
The 3 hardware servers with fibre channel are additionally set as proxies with "Max concurrent tasks: 12" each.
In addition to the 3 hardware proxies, we have a total of 16 virtual proxies (Linux and Windows mixed). The virtual proxies each deliver 4 "Max concurrent tasks" with 6vCPUs per virtual proxy.
In total, we have (16 virtual * 4 tasks) + (3 hardware * 12 tasks) = 64 + 36 = 100 concurrent tasks for the proxies. Approximately 1600 VMs are backed up.
After the upgrade to v11, however, we have extreme performance problems. The runtimes of both the backups and the copies are very long and no longer tolerable. The jobs stop with the message:
Code: Select all
20.03.2021 06:25:31 :: Waiting for backup infrastructure resources availability