I have a backup environment that consists of:
1 physical Backup Server with a 10gbit connection to the switch that is also the primary backup repository
1 VM for VMware backup proxy configured to use virtual appliance (8vCPUs & 16GB of RAM)
3 node vSAN cluster all connected to the same switch via 10gbit ports
When running backup jobs, Veeam starts running well, reaching read speeds close to 450MB/s for some jobs. After a while, the read speeds drops significantly for both incremental and full backups, going down to ~70MB/s in some cases.
The backup proxy and the backup repository are configured to process 8 concurrent tasks.
The bottleneck in all the backup jobs show 99% in source. How/where should I look for the reason of this bottleneck and what types of tests can I run to find which part is holding the backup speed down?