Refurbished storage for servers, NAS and SAN
Storage is the part of a server build where small details cause the most trouble. A drive with the right capacity but the wrong interface will never appear on the controller, and a perfectly healthy SSD is no use if its carrier doesn't fit the bay. This range covers the drives, the caddies that hold them and complete storage arrays, so everything a bay needs can arrive in one order. The notes below focus on interface, endurance and physical fit, because those decide whether refurbished solid state drives and hard drives work in production far more than headline capacity does.
At a glance
Typical deployments
Virtualisation datastores
Dozens of VMs reading and writing at once produce small random I/O, which is exactly what spinning disks handle worst. A handful of SSDs in RAID 10 will usually outperform a much larger set of 10K drives for this job. Choose mixed-use endurance where VMs write heavily, such as VDI or build agents, and read-intensive drives for steady application servers. Keep every drive in a datastore the same model and capacity. Our guide to servers for VMware and Hyper-V covers host sizing around the storage.
Databases and transaction logs
Latency and write endurance matter more than raw capacity. Keep transaction logs and temp space on their own SSD volume, separate from the data files, so sequential log writes aren't queued behind random reads. A busy log volume can wear a read-intensive SSD faster than expected, so check the drive's rated endurance against your daily write volume. Where the controller supports it, SAS or NVMe SSDs give deeper command queues than SATA.
File servers, NAS and backup
For a refurbished NAS server or backup repository, cost per terabyte wins and hard drive storage still makes the most sense. LFF 7.2K drives in RAID 6 give two-drive fault tolerance, which matters because rebuilding a large disk takes hours and stresses the rest of the set. Avoid RAID 5 on high-capacity HDD storage for that reason. If your NAS software supports it, a mirrored pair of SSDs for cache or metadata makes browsing and small-file work noticeably quicker.
Boot and operating system volumes
A hypervisor or operating system needs very little space and almost no write endurance. A mirrored pair of small SSDs, or 10K SAS drives where the bays are already SAS, is plenty. Don't put high-endurance drives here; save them for the volumes that actually write. Some servers can boot from internal modules instead, which frees the front bays for data.
ZFS, Ceph and software-defined storage
These platforms want to see each disk directly, so set the controller to HBA mode or fit a dedicated HBA rather than building hardware RAID underneath. Use the same drive model and firmware across a pool, and look for SSDs with power-loss protection for any log, journal or cache device. A ZFS storage server also uses spare memory as a read cache, so budget for RAM alongside the disks.
Shared SAN storage
When several hosts need the same block storage for clustering or live migration, refurbished SAN storage is often simpler than building replication in software. Check that the array's host connection (iSCSI, Fibre Channel or SAS) matches the adapters in your servers, look for dual controllers if uptime matters, and confirm which drive carriers the enclosure takes before buying disks to fill it.
If you are planning storage and servers together, pick the chassis first, because it decides the bay count, form factor and interface everything else has to match. Compare bay layouts across our refurbished servers, or spec a build with drives already fitted using the Server Configurator.
Why buy from Compuwyze
- 12-month return-to-base warranty on individual drives, caddies and other simple items, and 3 years return-to-base on configured servers and workstations.
- Orders for simple items placed before 2pm aim to ship the same day; configured servers aim to ship within 2 business days.
- Free UK delivery on orders over £50, normally arriving the next working day once shipped.
- International delivery with FedEx economy and priority services, usually within 2 to 5 days.
- Bulk drive orders and heavier items such as storage arrays shipped on pallets where needed, using couriers including Royal Mail, FedEx, FDC and DHL.
- 30-day returns, with return shipping paid by the customer.
Frequently asked questions
How much life is left in a refurbished SSD?
Enterprise SSDs are rated for a total write volume, usually expressed as drive writes per day (DWPD) or terabytes written (TBW), and they report how much of that has been used through SMART. Many server workloads write far less than the rating allows, so a drive with some endurance used can still be a sensible fit for a read-heavy role. Where a listing states a health figure, compare it with the write load you expect: a lightly worn read-intensive drive suits a boot or web volume, while a database log deserves more headroom.
Will a drive work in my server?
Check four things: form factor (SFF or LFF), interface (SAS, SATA or NVMe), what your controller and backplane support, and the carrier. Vendor-branded drives carry Dell, HPE or Lenovo part numbers, and some servers flag drives without the vendor's firmware as non-certified, so a matching part number is the safest choice. If you're replacing a failed disk, search by the part or FRU number on its label, or by model name such as Intel D3-S4510, to find a like-for-like match.
Do drives come with a caddy?
It depends on the listing, so check the product title and description before you order. If you need a carrier, choose the one listed for your exact server generation. Dell uses different carriers for 11th to 13th generation and 14th generation PowerEdge, so buy by generation, whereas HPE's standard SFF caddy (651687-001) is listed for Gen8, Gen9 and Gen10 ProLiant.
Can I mix different drives in one array?
Avoid it within a single RAID set if you can. A RAID volume runs at the pace of its slowest member and only uses the capacity of its smallest, and putting SSDs and hard drives in the same set wastes the SSDs entirely. Different models of the same capacity and interface will usually work together, but matching model and firmware gives the most predictable rebuilds. Put SSDs and HDDs in separate volumes on the same controller instead.
Should I choose a SAS or SATA SSD?
SATA SSDs cost less and are perfectly adequate for boot drives, read-heavy application data and many NAS roles. SAS SSDs add dual-port connections for redundant paths and deeper command queues, which suits busy databases and shared enclosures. NVMe is faster again, but only if the server has bays or slots wired for it. Your backplane decides first: a SATA-only backplane cannot run SAS drives at all.
How many spare drives should I keep?
For any array you rely on, keep at least one cold spare of the same model and capacity on the shelf, or configure a hot spare if you have a free bay. Drives bought together tend to age at a similar rate, so a spare lets you start a rebuild straight away rather than waiting for a replacement. For estates with several identical arrays, one spare per drive model is a sensible starting point.