Building resilient storage on a standalone server with Storage Spaces, and replacing a failed disk

Goal

Pool several physical disks on a single, standalone Windows Server host into a resilient virtual disk with Storage Spaces, and replace a failed physical disk without losing data or having to rebuild the pool from scratch.

Prerequisites

Enough physical disks of similar size for the chosen resiliency — on a standalone server a two-way mirror needs at least two, a three-way mirror at least five, single parity at least three and dual parity at least seven; local administrator rights; the Storage module.

Steps

1. Identify candidate disks not yet part of a pool: Get-PhysicalDisk -CanPool $true | Select-Object FriendlyName, SerialNumber, Size, HealthStatus.
2. Create the pool: New-StoragePool -FriendlyName "Pool1" -StorageSubSystemFriendlyName "*Storage*" -PhysicalDisks (Get-PhysicalDisk -CanPool $true).
3. Create a resilient virtual disk on top of it: New-VirtualDisk -StoragePoolFriendlyName "Pool1" -FriendlyName "Data1" -ResiliencySettingName Mirror -UseMaximumSize (use -ResiliencySettingName Parity for capacity-optimized, lower-write-performance resiliency instead).
4. Initialize, partition, and format the resulting disk exactly as any other disk (Initialize-Disk, New-Partition -UseMaximumSize, Format-Volume).
5. Monitor pool and disk health on an ongoing basis: Get-PhysicalDisk | Select-Object FriendlyName, HealthStatus, OperationalStatus and Get-StorageJob to see any in-progress repair or resync operation and its percent complete.
6. When a physical disk's HealthStatus reports Unhealthy or OperationalStatus reports a failure, identify it by serial number (identical models share a FriendlyName) and replace it in its enclosure, then add the new disk to the pool (Add-PhysicalDisk -StoragePoolFriendlyName "Pool1" -PhysicalDisks (Get-PhysicalDisk -SerialNumber <new>) -Usage Automatic), retire the failed one (Get-PhysicalDisk -SerialNumber <old> | Set-PhysicalDisk -Usage Retired), repair the affected virtual disks (Get-StoragePool -FriendlyName "Pool1" | Get-VirtualDisk | Repair-VirtualDisk -AsJob), and once Get-StorageJob shows the repair finished, remove the retired disk from the pool (Remove-PhysicalDisk -StoragePoolFriendlyName "Pool1" -PhysicalDisks (Get-PhysicalDisk -SerialNumber <old>)).

Expected result

Get-VirtualDisk shows HealthStatus: Healthy and OperationalStatus: OK once the repair job (visible via Get-StorageJob) completes; no data loss occurs for a Mirror or Parity virtual disk that lost only the number of disks its resiliency tolerates.

Limits and test basis

New-StoragePool, New-VirtualDisk, Get-PhysicalDisk, and Get-StorageJob are documented Storage module cmdlets; the disk-replacement sequence (replace and add the new disk, retire the old one, repair, then remove the old one from the pool) is documented in Microsoft's guidance on replacing failed disks and repairing JBODs for Storage Spaces. Repairing a large virtual disk after a full multi-terabyte drive failure can take many hours, during which the pool has reduced redundancy — avoid stacking a second planned disk removal on top of an in-progress repair. There is no single-command undo for a pool; back up data on it independently, since Storage Spaces resiliency protects against disk failure, not against deletion or corruption at the file level.