Storage usage of VM on NFS datastore is reported higher than the provisioned virtual disk size
Article ID: 436967
Updated On:
Products
Issue/Introduction
When checking the disk usage of a virtual machine on an NFS datastore using the ESXi command line (ls -isal "Used blocks") or the vSphere Client "VM > Summary > Usage > Storage", it may appear significantly higher than the provisioned (logical) size of the VMDKs.
For example:
13369482 2947751592 -rw------- 1 root root 2147483648000 mmm d hh:mm vm-name-flat.vmdk
Provisioned Size: 2,000 GB (2147483648000 byte)Physical Used (Blocks): Approximately 2.74 TB (2947751592 blocks * 1KB)
This discrepancy leads to the total directory consumption exceeding the sum of the logical disk sizes.
Environment
VMware ESXi
Cause
This behavior is by design of the NFS protocol and the underlying storage file system.
ESXi does not calculate the disk usage itself; it simply reports the number of blocks used as provided by the NFS server's backend.
The reported block count may include storage-side consumption that is invisible to the ESXi host, such as:
- Storage-side Snapshots:
Snapshots taken at the storage array level that retain blocks associated with the file. - Data Protection Overhead:
Redundancy or parity data that the storage array counts as part of the file's physical footprint.
Resolution
ESXi cannot distinguish why the backend storage is reporting a specific number of blocks.
To identify the exact cause of the consumption, you must inspect from the storage array management console.
For further investigation, please contact your storage hardware vendor.
Additional Information
Feedback
