Performance issue with GCP backup from snapshot
During GCP backup from snapshot operation the data is read from persistent disks attached to the Snapshot Manager. Persistent disk IOPS speed gets split between disks if read operation is going on multiple disks on the same machine.
For GCP backup from snapshot operation, a maximum number of 15 jobs can be launched (on machine whose capability is more than 15) and if the capability of the machine is less than 15, then those many backup from snapshot operation can run parallel on NetBackup Snapshot Manager.
If multiple backup from snapshot jobs are running, then . This results in longer backup times for the VM which have large size when large number of parallel backup jobs are going on.
Perform the following steps to improve the performance
- Select higher configuration for the NetBackup Snapshot Manager:
GCP disk IOPS depends on number of factors like VM type, Disk type, Disk size, CPU and so on.
Select higher configuration to get better IOPS. For more information, see Configure disks to meet performance requirements.
- Limit the number of jobs running on NetBackup Snapshot Manager:
Use the following settings in
/cloudpoint/flexsnap.conffile to limit the number of parallel jobs running on NetBackup Snapshot Manager:[capability_limit] max_jobs = <num>
If NetBackup Snapshot Manager machines capability is less than
max_jobsthen machines capability would be considered. If machines capability is more thanmax_jobsthen value ofmax_jobswould be used to decide the number of NetBackup Snapshot Manager jobs to be run on machine.For example, if max_jobs=8, then up to 8 GB of memory is available, which allows for 32 memory chunks (8 GB / 0.256 GB per chunk) and only 8 jobs, whichever is exhausted will stop scheduling new tasks
After changing the configuration restart the NetBackup Snapshot Manager and complete manual discovery on NetBackup.