A failed array isn’t dead.
It’s disassembled.
Multi-disk RAID recovery is a reconstruction problem, not a repair one. We clone every member, solve the stripe map, and rebuild the array in software. Nothing is written back to your original drives.
In-house lab in Huntsville. Free evaluation. No recovery, no charge.
Array state at intake
DegradedA single failed disk plus one weak drive is the most common case we see. Rebuilding without imaging first is where arrays die.
Why RAID is different
One drive isn’t the case. The relationship between drives is the case.
In a single-drive recovery we pull data off a failed device. In a RAID case we have to rebuild how the data was spread across the devices before we can pull anything. Miss the stripe size by one block and the rebuilt array is noise.
What has to be solved before any file opens
Which physical disk was at which array position
Block size used when writing across disks (4K to 1 MB)
Left-symmetric, right-asymmetric, or forward
Where data begins on each member relative to the disk
Failure by level
How each RAID level breaks, and how we bring it back.
One failed disk means the stripe is broken. We image every remaining member, then reassemble stripe order and block size to pull coherent data.
Mirrored sets look simple until one side silently diverges. We compare both mirrors sector by sector and pick the most recent valid copy.
Single-disk failure is survivable. Double-disk failure or a weak drive during rebuild is where most cases land. We solve parity direction and rotation to rebuild virtually.
Tolerates two failures but rebuilds are long and read-heavy. When a third drive shows reallocated sectors mid-rebuild, the array locks. We clone first, rebuild second.
Two failures in the same mirror pair end the array. We identify which pair is broken, rebuild the stripe from the surviving mirror, and validate.
Metadata on the drives defines the array. Lost or corrupted superblocks are reconstructed from member headers and data patterns.
Do not rebuild a degraded array
Most RAID data loss happens during the rebuild, not the original failure.
A rebuild reads every sector
On a degraded array, one weak sector on a remaining drive during rebuild can lock the whole set. Imaging first preserves every member.
Parity can be written wrong
If the controller misidentifies member order after a failure, a rebuild overwrites good data with incorrect parity. The clones keep the original untouched.
Rebuilds stress weak drives
Member drives that were marginal before the failure often die during the rebuild itself. We clone in write-protect mode, so the original drives are not asked to do it again.
Our bench
Every member drive stays on our rack, from intake to return.
RAID cases never leave our Huntsville facility. Imaging, parity analysis, filesystem work, and verification all happen on the same bench, by the same team.
Parity analysis, stripe detection, virtual rebuild
Up to 12 members imaged in parallel with write-protect
Head swaps for member drives with mechanical failure
NTFS, ReFS, ext4, XFS, APFS, HFS+, VMFS
How we treat every array
- 1
Originals are write-protected before power is ever applied.
- 2
Every member is cloned to a dedicated image target. We never work off the originals.
- 3
Parity geometry is solved from the image data, not trusted from controller metadata.
- 4
Virtual rebuild runs against the clones. If a pass fails, we iterate; the originals stay untouched.
- 5
Filesystem mount is always read-only. We verify files before anyone signs off.
Pricing
Scoped per array. Firm quote before work.
Evaluation
Free
No deposit to begin
- Drive-by-drive health check
- Array metadata and parity readability
- Recovery feasibility and time estimate
- Firm written quote before any work
Logical array rebuild
From $295
All members healthy, metadata intact
- Lost superblocks or RAID metadata
- Accidental reinitialization or rebuild misfire
- Filesystem damage on a healthy array
- Typical turnaround 2 to 5 business days
Multi-disk & controller
From $495
Failed disks, dead controller, bad rebuild
- Sector-accurate imaging of every member
- Virtual rebuild with solved stripe map
- Head swap or donor work on member drives
- Scales by number of member drives
Final cost scales with member count and failure type. Cases with head swaps or donor boards are quoted after evaluation.
Recovery flow
Four stages. One array. No guesswork.
Intake and labeling
Every member drive is logged by bay position, serial, and health state. Originals are set to write-protect before anything else.
Clone and map
Each member is cloned to a dedicated image target. We read the metadata and solve stripe size, rotation, and parity direction from the image data.
Virtual rebuild
The array is reconstructed in software from the clones. Original drives are never used for the rebuild itself.
Verify and return
Filesystem is mounted read-only, file checksums are confirmed, and data is delivered on an encrypted drive or secure upload. Member drives are returned.
Supported systems
Controllers and filesystems we work with
Controllers
- Dell PERC (H700 / H730 / H740)
- HP Smart Array (P4xx / P8xx)
- LSI / Broadcom MegaRAID
- Adaptec Series 7 / 8
- Areca ARC series
- Software (mdadm, Storage Spaces)
Filesystems
- NTFS, ReFS
- ext4, XFS, Btrfs
- APFS, HFS+
- VMFS (ESXi)
- ZFS pools
- exFAT / FAT32
Array not listed? Call us. If it stripes or mirrors, we’ve probably seen it.