Coordinate every member disk when changing Write Accelerator on a striped log volume

Source facts

Write Accelerator is supported for Premium SSD managed disks on M-series VMs in Azure public cloud. Microsoft recommends it for database transaction or redo logs, not database data volumes. When an existing volume is striped across several Premium SSDs, every member must be enabled or disabled in separate steps, with the VM shut down first. Supported cache modes are None or Read Only. Accelerator IOPS limits are shared across the VM’s accelerated disks, not granted separately to each disk. Microsoft Learn.

Applicability

Review this procedure for an existing striped log volume, including the documented Windows volume-management, Storage Spaces, SOFS, Linux LVM, or MDADM arrangements. Identify every underlying disk and the actual application dependency before scheduling the change.

DSE recommendation

Approve the complete striped-volume change and VM shutdown before altering any member disk. Have the storage and database owners agree on the member list, desired setting, cache mode, and post-start checks. Keep the aggregate accelerator limit in the performance plan so adding members is not mistaken for multiplying that allowance.

Verification

After the authorized change and start, compare the setting across every member disk and verify the volume is available to the database. Exercise the intended log workload and record observed behavior against its acceptance criteria. Investigate a mixed configuration or missing member before returning the change to normal operations. Preserve the full mapping and actual results; a successful update to one disk does not establish completion for the striped volume.

Official references

Microsoft Learn: Enable Write Accelerator. Source reviewed September 9, 2026.