How Do You Order the Right Replacement for a Failed RAID Drive?

When a drive in a RAID array fails, the replacement has one job: be accepted by the controller so the rebuild starts and finishes cleanly. That comes down to matching five attributes of the failed drive: interface, form factor, capacity, spindle speed and sector format. The brand on the label is not one of them. The sixth thing, the drive carrier or tray, is the one most often forgotten until the box arrives. This guide covers how to read those attributes, what you can safely change, and how to swap the drive without putting the array at risk.

What replacement drive do I need for my RAID array?

Start with the controller, not the drive label. The server's management controller (iDRAC, iLO or XClarity) or the RAID controller's own utility reports the failed drive's model, capacity, interface and negotiated link speed. The label on the drive itself gives you the manufacturer or OEM part number and, usually, the sector format. With those in hand, match the following:

• Interface: SAS for SAS, SATA for SATA. Many controllers refuse to mix SAS and SATA drives in the same virtual disk, even when the backplane accepts both.
• Interface speed: a 12Gb/s SAS drive works in a 6Gb/s SAS backplane and simply runs at 6Gb/s. The link always negotiates to the slower side, so speed is rarely a blocker.
• Form factor: 3.5-inch LFF or 2.5-inch SFF. Order the size the bay was built for, in the carrier that fits it.
• Capacity: equal or larger, never smaller. A controller will not accept a smaller drive as a rebuild target.
• Spindle speed: 7200 RPM for 7200 RPM, 10K for 10K. A slower drive, if the controller accepts it at all, drags the whole array down to its pace.
• Sector format: keep the logical sector size consistent with the other members. 512n and 512e both present 512-byte sectors; 4Kn does not, and cannot join a 512-byte array.

A worked example: the failed drive is a 4TB 7200 RPM 12Gb/s SAS 512e LFF drive in a PowerEdge. The direct match is the Dell 4TB nearline SAS drive, which is dual-port, CMR and ships in the Dell hot-plug carrier. The equivalents for the other two big OEM platforms are the HPE 4TB midline SAS drive with the HPE Smart Carrier, qualified for ProLiant Gen9 to Gen11, and the Lenovo ThinkSystem 4TB with its hot-swap tray for ThinkSystem SR and ST servers. Same capacity, interface, speed and sector format in each case; the difference is the carrier and the platform qualification.

Does the replacement drive have to be the same brand?

No. RAID controllers track members by interface, capacity, block size and health. An array built from Seagate drives rebuilds onto a Western Digital or Toshiba drive of the same specification, and plenty of arrays end up mixed after a few years of replacements. If you are replacing a 12TB SAS member behind a standard controller, the Seagate Exos X18 12TB (7200 RPM, 12Gb/s SAS, 512e, LFF) is a spec match whatever brand the original was.

The brand question that does matter is OEM-branded versus manufacturer-branded. Dell, HPE and Lenovo servers are built around drives that carry their firmware and fit their carriers. A generic drive will often work, but it arrives without a tray and may raise warnings in the management controller. If the server is under OEM support, or you want the drive to drop straight into the bay, order the OEM part with its carrier.

Can I replace a failed drive with a bigger one?

Yes, and sometimes you have to, because the original capacity is no longer made. A larger drive is accepted as a rebuild target, but a conventional RAID set only uses as much of each member as its smallest drive provides. Put a 22TB drive into an array of 12TB drives and the array still sees 12TB on that slot. The extra capacity becomes usable only after every member has been replaced and the virtual disk and file system are expanded, and only if your controller supports online capacity expansion.

Two checks before going bigger. First, recording technology: stay with CMR (conventional magnetic recording) for RAID members. The Seagate Exos X22 22TB, WD Ultrastar DC HC570 22TB and Toshiba MG11 22TB are all CMR, dual-port 12Gb/s SAS, 512e LFF drives, which makes them interchangeable members of the same array. Second, rebuild time grows with capacity. A large nearline drive takes much longer to rebuild than the 4TB drive it replaces, and the array is exposed for that whole window.

What if the array was built on older 512n drives?

Older platforms often run 6Gb/s SAS drives with 512n (native 512-byte) sectors. For those, replace like with like where possible. The HP 2TB 6Gb/s SAS 512n drive comes in an HP Smart Carrier for ProLiant G6 to Gen8, the IBM 2TB 512n nearline drive covers System x M4 and M5, and the Dell EMC 4TB 512n drive is built for Unity and VNX arrays with its carrier included. A 512e drive also presents 512-byte logical sectors, so many server RAID controllers accept it in a 512n array, but storage arrays such as Unity and VNX expect their own qualified parts, and a matched drive removes the question entirely.

How do I swap the drive without risking the array?

• Confirm the degraded array and the failed slot in the controller log, then switch on that slot's locate LED. Pulling the wrong drive from a degraded RAID 5 array takes it offline.
• Check that the last backup completed. A degraded array has lost its redundancy (or, in RAID 6, part of it) until the rebuild finishes.
• Look at the other members' health and media error counts. Drives bought together tend to age together, and a rebuild reads every sector of every surviving drive.
• Insert the replacement in the same slot. Most controllers start the rebuild automatically on a hot-swap bay; some need the drive marked as a hot spare or the rebuild started by hand.
• Keep the workload light during the rebuild where you can, and confirm the virtual disk returns to an optimal state before closing the ticket.

Match interface, form factor, spindle speed and sector format, at equal or larger capacity. The brand is the one attribute that does not need to match.

Should I order a spare while I am at it?

Yes. If one drive from a batch has failed, the others share its age and workload. Keep one or two cold spares of the same specification on the shelf, or configure a global hot spare so the next rebuild starts without waiting for a delivery. For arrays built on large nearline drives, it is also worth asking whether RAID 6 or RAID 10 is a better fit than RAID 5 at the next refresh; our guide to RAID levels for large datasets covers the trade-offs, and RAID vs HBA covers the controller side. Performance tiers built on 2.5-inch 10K RPM SAS drives follow exactly the same matching rules.

Nexus Compute supplies new drives sourced through authorized distribution with the full manufacturer warranty. Send us the part number from the failed drive's label or the controller log, and we will confirm the match against your server or array before dispatch. Pricing is on each product page, or request a quote for multiple drives and spares, answered within 48 business hours.