Skip to main content
commvault.com commvault.com

we have vmware vms hosted on HPE Alletra 9000 storage. Small vms are backed up using snapshot of the vms . The storage policy uses a global dedupe storage pool, so the current backups are copied to Hyperscale 1.5 storage libraries. the local snapshots disappear when the job completes.

I understand that intellisnap backups can use the snapshot ability of the allatra storage array. 
We currently use multiple linux based physical hyperscale nodes as mediaagents which already have the mediaagent and vsa software built in.
I expect that separate commvault subclients have to be setup in vcenter > vmware for each datastore that we would like to use an intellisnap backup? We have the alletra arrays configured in array management with all the mediaagents as access nodes
My questions are:
1) would  the sub clients would use the HPE 3par snap engine? instead of the vcenter
2) Some datastores host several large vms, how can a subclient that points to that datastore process such a large alletra datastore snapshot if the mediagents are also busy doing normal vmware backup using the current method? Should each subclient have a different mediaagent specified as the vsa proxy? 
3) how many datastores can have an intellisnap snapshot taken at the same time? Like do we have to stagger the backup and processing of the backup copy to avoid lockups and bottlenecks ?        

If you can align the VMs you are backing up with datastore affinity, that is a much cleaner setup for IntelliSnap. It means during backup copy operations you are only mounting one snapshot, not one snapshot per datastore, so I would recommend it if possible.

  1. Yes it looks like the 3par engine supports Alletra 9000
  2. There is real no processing on an intellisnap backup. The VMs are snapped within VMware (software snapshot) to capture a consistent copy of the VM before the underlying hardware snapshot is done to capture the VMs on the datastore. Once the hardware snapshot is done, the software snapshot is released. This should be less impactful than a software snapshot that has to wait around until the data is streamed out. Do not use the ‘run backup copy immediately’ option as you loose a lot of control of the environment that way. Rather schedule backup copy jobs separately.
  3. There is no way to tell, its entirely dependent on your environment. You can enable client throttling and then create your subclients based on datastores. This way you can have control over the number of datastores snapped. E.g 10 subclients assigned to one datastore each. Limit to 5 jobs = 5 datastores snapped concurrently

For backup copy operations, I’d recommend using the proxyless option where possible: https://documentation.commvault.com/2023e/expert/intellisnap_backup_operations_and_backup_copy_operations_run_without_proxy_esxi_host.html

Reply


Terms & ConditionsCookie settings