Skip to main content
Question

SQL hosted on VM - Instant Clone / Snapshot


Forum|alt.badge.img

Hi,

I have a 20TB SQL VM that I would like to enable Intellisnap for quick clone and quick restores.

 

The issue is that it uses regular VMDK on a Hitachi VSP and the VMware team is not willing to change it pRDM.

 

Is there any way I could still use Intellisnap for SQL with VMDks?

 

I saw this demo for IBM copy data management + powerflex and I was wondering if Commvault could do the same with Hitachi 

 

2 replies


Forum|alt.badge.img
  • Author
  • Apprentice
  • October 23, 2026

As far as I understand, with SQL hosted on a VMDK I can only use Intellisnap for the VM and not the SQL level.

This mean I can quickly clone the Full VM using Live Recovery from Intellisnap.

SQL IntelliSnap clone and fast database restore effectively require pRDM. Many VMware teams refuse pRDM for operational reasons such as vMotion, standardization and supportability.

Large SQL VMs on VMDK (ours is 20TB on Hitachi VSP) can only use VM-level IntelliSnap. That gives us VM or VMDK restores, but no database-centric instant clone, instant restore or point-in-time recovery workflow.

Requested capability

A SQL-aware IntelliSnap workflow for databases on VMDKs on VMFS datastores, starting with Hitachi VSP through Thin Image and the CCI engine:

Discovery: automatically map SQL instance → databases → VMDKs → datastores → VSP LDEVs.

Consistent snapshot: quiesce SQL in the guest (VSS), take the Thin Image snapshot of only the datastore LDEVs backing that database, then release.

Instant clone: present the snapshot to a target ESXi host and resignature or mount the datastore. Attach the VMDKs to a chosen target VM and attach the databases under a new name, all in one job, in minutes, with no data copy.

Instant restore: the same mount-and-attach flow to restore a database in place or side by side, from any snapshot.

Point-in-time restore and clone with logs:

Run SQL transaction log backups through the in-guest SQL agent on their own schedule, alongside the snapshots, for example every 15 minutes. Catalogue them against the snapshot chain.

When restoring or cloning, let the user pick a date and time, or an LSN or marked transaction, instead of a snapshot.

Commvault automatically mounts the nearest snapshot before that point and attaches the database in NORECOVERY/STANDBY mode. It replays the required log backups up to the exact time requested, then brings the database online.

Support in-place restores, side-by-side restores (new database name), and clones to a different target VM.

Validate the log chain before the job starts, and report any gaps along with the earliest and latest recoverable points.

Lifecycle: scheduled clone refresh, expiry, and automatic cleanup of the mounted datastores and snapshots.

Business value

Dev/test and reporting copies of a 20TB database in minutes instead of hours.

RTO measured in minutes for large databases.

Snapshot-speed recovery combined with near-zero RPO, so we can recover from logical corruption or a bad deployment to the minute before it happened.

Snapshot-based protection without forcing infrastructure changes such as pRDM on the VMware team.