Buying a second hand SSD sensibly comes down to three things: the interface your backplane or motherboard actually presents, the capacity you need, and how much write endurance the workload will consume over its remaining life. Get those right and a refurbished enterprise drive gives you flash-level latency for a fraction of the cost of new — with the same tested performance behind it.
A 2.5in SATA or SAS SSD fits a standard 2.5in server caddy, so check you have the right carrier for your chassis before ordering.
At a glance
- Form factors
- 2.5" (7mm and 15mm), M.2 2280, and NVMe add-in-card / U.2 style drives for servers and workstations
- Interfaces
- SATA 6Gb/s, SAS 12Gb/s and PCIe NVMe — the listing states the interface for each drive
- Capacities
- From boot and OS-class drives around 256GB through to large-capacity data and archive tiers
- Brands stocked
- Samsung, Intel and Micron, alongside OEM-branded drives carried by server vendors
- Endurance classes
- Read-intensive, mixed-use and write-intensive — quoted by the manufacturer in DWPD or total TBW
- Testing
- Every drive is tested for performance and endurance before dispatch, then graded
- Warranty
- 12-month return-to-base cover as standard
Typical deployments
Virtualisation hosts
Random I/O from dozens of VMs is the challenge, not sequential throughput. Prioritise mixed-use endurance and consistent low latency under queue depth over headline sequential figures, and favour more smaller drives over one large one so a single failure costs you less rebuild time.
Databases and OLTP
Write amplification from logs, indexes and checkpoints eats endurance quickly. Write-intensive drives with power-loss protection earn their premium here; NVMe helps most where you are latency-bound rather than bandwidth-bound, which is usually the case for transactional work.
Boot and hypervisor drives
A small read-intensive SATA or M.2 drive — 256GB is plenty for most hypervisors — mirrored for redundancy. Endurance barely matters; what matters is that the drive is supported by the boot controller or M.2 riser in your chassis.
VDI and remote desktop
Boot storms and non-persistent desktops produce heavy bursts of writes against a shared pool. Size for concurrent IOPS per user rather than raw capacity, and keep some overprovisioned space free so garbage collection does not collide with peak login periods.
Caching and tiering
Read cache in front of spinning disk is the cheapest way to lift an existing array. Read-intensive drives are fine for a read cache; if the tier absorbs writes as well, step up to mixed-use so the cache is not the first thing to wear out.
Workstations and media
Video editing, rendering and CAD scratch volumes reward sequential bandwidth and sustained write performance. An NVMe M.2 or add-in-card drive gives the best return, with a larger SATA SSD behind it for project storage.
Why buy from CompuWyze
- Performance and endurance tested before dispatch, with each drive graded so you know what you are getting
- 12-month return-to-base warranty on every SSD we ship
- Enterprise SATA, SAS and NVMe held side by side, so you can match the drive to the backplane you actually have
- Recognised flash from Samsung, Intel and Micron rather than unbranded consumer stock
- Capacities from 256GB boot drives upwards, useful when you need to fill bays consistently
- Part numbers listed so you can cross-check against your server vendor's compatibility documentation
Frequently asked questions
What does refurbished mean for an SSD?
These are used, pre-owned enterprise drives that have been pulled from service, wiped, and then tested for both performance and endurance before being graded and put on the shelf. Because flash wears with writes, endurance testing matters as much as a functional check — a drive that passes a simple read test can still have consumed most of its rated write life.
Can I use a second hand SSD in production?
Yes — enterprise SSDs are built for duty cycles far beyond what most deployments impose, which is why refurbished drives are common in virtualisation hosts, database servers and high-throughput storage. Match the endurance class to the workload, keep drives in a redundant array rather than standalone, and monitor SMART wear indicators as you would with new flash.
Are caddies, trays or brackets included?
Only where the individual listing says so. Bare 2.5" drives normally ship without a vendor-specific hot-swap carrier, and M.2 or add-in-card drives may need a riser or adapter depending on the chassis. Check the listing and ask us if you are unsure which carrier your server takes.
SATA, SAS or NVMe — which should I choose?
It is usually decided for you by the backplane and controller. SATA is the broadest fit and the cheapest per gigabyte; SAS adds dual-porting and generally higher sustained performance on a SAS HBA or RAID controller; NVMe bypasses the storage controller entirely for the lowest latency, but needs PCIe lanes and an NVMe-capable slot, riser or U.2 backplane. Confirm what your server presents before ordering.
Is any previous data left on the drives?
No. Drives are wiped as part of processing, and arrive ready to be initialised by your operating system, RAID controller or storage software.
Can I mix these drives with the ones already in my array?
Generally you want identical capacity and, ideally, the same model and firmware within a RAID group or vSAN disk group — mixed drives work but the array is limited by the slowest member, and some controllers are fussier than others. If you are expanding an existing pool, tell us the part number you are matching and we will point you at what we have.
Why buy a small 256GB SSD rather than something larger?
Boot and hypervisor volumes do not need capacity, they need reliability and a supported interface. A pair of small read-intensive drives mirrored for the OS keeps your larger, higher-endurance flash free for data — and costs very little to standardise across a fleet.