Calculate usable capacity, fault tolerance, performance (IOPS), rebuild time, and failure risk for any RAID configuration. Built for IT teams planning refurbished servers and workstations.
Estimated random 4K I/O for the chosen drive type. Large sequential writes run faster — full-stripe writes avoid the parity read-modify-write penalty.
Build this configuration with PC Server & Parts
RAID (Redundant Array of Independent Disks) combines multiple drives into one logical volume to gain capacity, speed, redundancy — or a balance of all three. Here is how the common levels compare.
Striping across all drives for maximum speed and full capacity — but a single drive failure loses everything.
Min 2 drives · ~100% usable
An exact copy of your data on two drives. Simple, reliable redundancy for boot/OS volumes — for more drives, use RAID 10 instead.
2 drives · ~50% usable
Striping with one parity drive's worth of redundancy. Good capacity efficiency, but risky to rebuild on large drives.
Min 3 drives · ~67–94% usable
Two parity drives survive two simultaneous failures — and a read error during a rebuild. The safe default for large HDD arrays.
Min 4 drives · ~50–88% usable
Striped mirrors: the performance of RAID 0 with the redundancy of RAID 1. The go-to for databases and VMs.
Min 4 drives · ~50% usable
Multiple RAID 5 groups striped together — better performance and rebuild behavior than one wide RAID 5.
Min 6 drives · ~67–94% usable
Multiple RAID 6 groups striped together — high capacity with double-parity safety per group. For large bulk storage.
Min 8 drives · ~50–88% usable
Match your top priority to a RAID level — then size the drives and chassis with the calculator above.
Whatever you choose: add a hot spare, use a dedicated RAID controller where you can, and remember RAID is not a substitute for backups. New to this? What is a RAID array? compares every level side by side, including hardware versus software RAID and what each one costs in usable capacity.
The calculator estimates rebuild time from three inputs: the failed drive's capacity, its sequential speed, and a rebuild-efficiency factor that accounts for the controller rebuilding while the array keeps serving normal I/O.
rebuild time ≈ drive capacity ÷ (sequential speed × rebuild factor)
Example — recovering a failed 8 TB SATA HDD in a RAID 5 array: 8 TB ÷ (180 MB/s × 0.4) ≈ 31 hours. RAID 5 recovery reads every surviving drive end-to-end, so a wider array doesn't rebuild faster — the failed drive's capacity sets the floor. Until the rebuild finishes the array runs degraded, and one more failure means data loss.
| Failed drive | Est. rebuild time |
|---|---|
| 4 TB SATA HDD | ~15 hours |
| 8 TB SATA HDD | ~31 hours |
| 12 TB SATA HDD | ~46 hours (≈2 days) |
| 8 TB enterprise SAS HDD | ~18 hours |
| 16 TB enterprise SAS HDD | ~36 hours |
| 1.92 TB SATA SSD | ~1.7 hours |
| 3.84 TB NVMe SSD | ~30 minutes |
Real-world times vary with the controller's rebuild-priority setting and production load — a busy array can take 2–3× longer. RAID 10 is the exception: it copies a single mirror instead of reading the whole array, so rebuilds are fast and low-stress. Set your exact drives in the calculator above to get the estimate for your configuration.
For parity RAID the danger isn't only a second dead drive — it's hitting an unrecoverable read error (URE) while the rebuild reads every surviving drive. On RAID 5 a single URE aborts the rebuild and costs data. The calculator estimates that probability from the drive's rated error rate and the amount of data the rebuild must read:
P(rebuild failure) = 1 − e−(bits read × URE rate)
| Drive class | URE rate |
|---|---|
| Desktop SATA HDD | 1 per 1014 bits (≈12.5 TB read) |
| Enterprise SAS HDD | 1 per 1015 bits (≈125 TB read) |
| SATA SSD | 1 per 1016 bits |
| NVMe SSD | 1 per 1017 bits |
The numbers get dramatic fast: rebuilding a 4 × 8 TB RAID 5 on desktop-class SATA drives reads ~24 TB and has an ~85% chance of hitting a URE mid-rebuild. The same array on enterprise SAS drives drops to ~18% — and on RAID 6 a read error is auto-corrected by the second parity, so the rebuild simply continues. That's the whole argument for RAID 6 on large HDD arrays. New to RAID concepts? Start with our plain-English guide: what is RAID?
RAID math is vendor-neutral, so this works as a Dell RAID calculator for PowerEdge PERC controllers (H710, H730, H740P, H755), for HPE ProLiant Smart Array (P408i, P816i), for Broadcom/LSI MegaRAID cards, and for software RAID like ZFS, mdadm and Windows Storage Spaces. Usable capacity, fault tolerance and the rebuild window are identical across vendors; only effective IOPS shifts with the controller's cache size and write-back battery. Need the hardware? Shop RAID controller cards or refurbished servers with the controller pre-installed. And remember every hot-swap drive needs its own caddy — the Drive Tray & Caddy Finder lists the exact part numbers for your server model.
