Interface first, then what makes it different
Every kind of member drive, and how each one fails. The connector tells us the bench; the label tells us the rest.
The drive that comes out of a server slot is not the drive that goes into a desktop. It speaks SAS or SCSI rather than SATA, or it speaks SATA with enterprise firmware that gives up on a bad sector in seconds so the controller can carry on; it may have 520 bytes in every sector, a key that only its controller holds, helium sealed inside it, or a vendor's firmware that refuses any other host. What decides the job is the interface, the format, the security state and the mechanics, in that order. Pick yours below. One failed member drive, imaged so the set can be rebuilt, is £300 + VAT; a set of two to four members is £500 + VAT upwards; larger sets, 15K SAS sets and storage shelves start at £1,250 + VAT, with the figure in writing after the free look and no data meaning no bill on most jobs.
Rather talk it through? An engineer answers the bench line
0800 6890668
By the interface on the back, or by what makes it different.
Rather begin with what it says →By the interface on the back
SAS 2.5-inch 10K and 15KST…MM and MP, HUC10K and 15K, AL14 and AL15: the fast drives in 1U and 2U servers. Heads first, then firmware. £300 + VAT→NL-SAS 3.5-inchExos and Ultrastar on a SAS board: the big drives in MD1200, D2600 and DS4246 shelves. Often SED, often 4Kn→Enterprise SATA 3.5-inchExos, Ultrastar, Toshiba MG, WD Gold. Time-limited error recovery means the controller drops them fast→SCSI and Fibre ChannelCheetah, Atlas, Fujitsu MAW: 68-pin, 80-pin SCA and 40-pin FC drives from PowerEdge 2850s and old SANs. Terminated, then imaged→SAS and SATA SSD membersHPE VO/MO/EO, Dell LT, Samsung PM893, Micron 5300. Controllers and firmware, sometimes on a schedule→NVMe U.2, U.3 and M.2 membersPM9A3, P4510, CM6, SN840, Micron 7450. A PCIe adaptor for the form factor, then imaging→By what makes it different
NAS drives as membersIronWolf, WD Red and N300 pulled from a Synology, QNAP or TrueNAS box. SMR translators and 10TB timeouts→Surveillance drivesSkyHawk and Purple from NVRs, written to around the clock. Stop the recorder before the loop overwrites the week→520-byte and T10 PI drivesNetApp, EMC, Sun and IBM storage drives a standard host reports as 0 GB. Imaged at native size; the drive is never reformatted→Self-encrypting drivesTCG Enterprise and Opal drives locked to a controller. Unlocked with your key; a PSID revert ends everything→Helium and SMR membersHe8 to HC590, Exos X10 onwards, MG07 onwards: welded shut. WD Red EFAX and HC6xx: shingled. Different benches→Dell, HPE and Lenovo-brandedA Seagate, WD or Toshiba in a carrier with vendor firmware. Dell D-revisions, HPE HPD-revisions, and the clocks inside them→Why the interface matters, and why the set matters more.
A desktop hard drive answers a bad sector by trying, and trying, for as long as it takes. An enterprise drive gives up in seconds, because a RAID controller cannot wait: it drops the drive from the set, marks it failed, and carries on from the others. That is why so many drives arrive here marked failed by a controller and still perfectly readable on a bench that is prepared to wait. It is also why the set matters more than the drive. The drive the controller dropped is usually recoverable; the question is what the rest of the set has been asked to do since.
Every drive on this page is imaged on the equipment its interface needs, at its native sector size, with the weak areas last, and never in a server. If one member failed and the set is otherwise healthy, the image goes back on fresh media, sector-identical, so the rebuild can run from a drive that answers every read. If the set itself has failed, every member is imaged and the array is reassembled from the images in software, which is the work our RAID array site describes and the same bench does. There is more on both on What is inside a RAID member drive, and a page for each family and host on the manufacturers page.
Two kinds of drive ask more. Drives from NetApp, EMC, Sun and IBM storage use 520 or 528 bytes in every sector and are reported by an ordinary host as 0 GB or incompatible; they are imaged as they are and the extra bytes stripped from the image. Self-encrypting drives, which most SAS and many NL-SAS drives now are, hold their data behind a key on the controller; they are unlocked on the imager with the key you supply, and without it there is nothing to recover, which we say at the free look.
The questions that come up first.
Can I put a SAS drive in my PC to check it?
Not without a SAS host adaptor, and not at all if the data matters. A SAS drive is imaged on a SAS imager that controls every retry; a PC that cannot see it tells you nothing, and one that can will try to initialise it.
Do I send one drive or the whole set?
If one member failed and the set is otherwise healthy, send that drive and it comes back imaged onto fresh media for the rebuild. If a second drive is flagged, or the set has failed, send every member labelled by slot.
The warranty engineer is coming for the failed drive. Should I hand it over?
Not until you have decided you do not need its data. A drive that leaves with the courier is a drive that leaves. Dell's Keep Your Hard Drive and HPE's retention options let you keep it; ask, and send it here first.
Does the interface change the price?
No. The number of drives does. One member is £300 + VAT; two to four are £500 + VAT upwards; larger sets, 15K SAS sets and storage shelves from £1,250 + VAT. Fixed in writing after the free look.
Not sure which one you have?
Photograph the label and the connector and send them with the form. The model string is enough to tell, and the first look is free whichever kind it turns out to be.