VM storage consumption increases after changing vSAN policy from RAID-5 to RAID-1 - VMware vSAN
Article ID: 448882
Updated On:
Products
Issue/Introduction
When a virtual machine storage policy is updated from RAID-5 (Erasure Coding) to RAID-1 (Mirroring) on a vSAN Original Storage Architecture (OSA) datastore, the physical storage footprint of the VM increases. This behavior is expected due to the 2x overhead required for mirroring. This growth can lead to cross-vCenter migration failures if the destination cluster lacks sufficient physical capacity.
Symptoms
- A thin-provisioned VMDK grows immediately after a policy change (e.g., a 29TB disk growing to 48.5TB).
- vCenter reports "Used Space" nearly double the logical data size within the Guest OS.
- Cross-vCenter migrations fail with out-of-space errors at the destination due to unexpected physical disk growth.
- Object re-syncing tasks are observed in the vSphere Client following the policy update.
Environment
- VMware vSAN Original Storage Architecture (OSA)
- vSAN Datastore
- Cross-vCenter Migration
Cause
The increase in storage consumption is due to the difference in physical overhead between RAID-5 and RAID-1 policies in vSAN OSA:
When a policy is changed to RAID-1, vSAN must allocate physical blocks for the mirror components. For a 29TB VM, the physical usage will trend toward 58TB as vSAN mirrors all "dirty" or written blocks.
Resolution
To reduce the physical storage footprint and return to the original logical size, follow these steps:
- Delete VM Snapshots: Ensure all VM snapshots are deleted before changing the storage policy. Snapshots can prevent full space reclamation and block the successful application of a new policy.
- Revert to RAID-5 Storage Policy:
- Navigate to the VM in vCenter.
- Select VM Policies > Edit VM Storage Policies.
- Re-apply a RAID-5 (Erasure Coding) policy. This triggers an object resync to release mirror components.
- Guest OS UNMAP:
Feedback
