Replace one drive at a time, letting the array repair fully between swaps. In RAID 1, 5, 6, and 10 capacity expands only after the last drive is replaced. At today's live prices the smartest jump for a 4-bay RAID 5 array upgrading from 8TB is 20TB, at about $26.89-28.78 per added usable TB after reselling the old drives.
Take a full, verified backup first. This is not boilerplate: RAID 5 runs degraded during every single rebuild, and an upgrade means back-to-back rebuilds, so the days you spend swapping drives are statistically the riskiest in your array's life. A backup you have actually tested restoring from is the difference between an inconvenience and a disaster if a second drive drops mid-rebuild.
Then confirm the pool is Healthy before you begin — never start an upgrade on an array that is already degraded or showing SMART warnings. Check your NAS vendor's drive compatibility list for the capacity and model you plan to buy. And buy all the replacement drives up front. In the current shortage, a mid-swap stock-out can leave you stranded with a half-migrated array and no way to finish with matching drives at the same price — purchasing the full set at once removes that risk and lets you complete the sequence quickly.
Platform notes. On Synology, the flow above matches the Storage Manager repair workflow, and RAID 1/5/6/10 and F1 expand automatically after the final swap; SHR benefits earlier when you mix in larger drives. QNAP's QTS follows the same one-by-one-and-repair pattern. UnRAID is different: it is not parity-striped like traditional RAID, so the parity drive must be the largest in the array — replace parity first if you are also raising the array's maximum drive size (see our UnRAID guide).
This is the part almost no guide covers, because it needs live prices. The right capacity to jump to is not simply the biggest you can afford — it is the one with the best net cost per added usable terabyte after you resell the drives you pull. The table below runs that calculation live for a 4-bay RAID 5 array upgrading from 8TB drives: for each candidate capacity it shows the cheapest in-stock drive, its raw dollars per terabyte, and the net dollars per added usable terabyte after crediting the resale of your four old 8TB drives (each worth $51-68 today, anchored to a $85 refurb price).
| Jump to | Cheapest in-stock | $/TB | Net $/added usable TB |
|---|---|---|---|
| 20TBBEST | New $310 · MDD 20TB SAS 12G 7200RPM Enterprise | $15.50 | $26.89-28.78 |
| 24TB | Refurb $399 · WD Elements Desktop 24TB USB | $16.63 | $27.58-29.00 |
| 22TB | New $380 · MDD 22TB SAS 12G 7200RPM Enterprise | $17.27 | $29.71-31.33 |
| 18TB | New $309 · WD Ultrastar DC HC550 18TB SATA | $17.17 | $32.13-34.40 |
| 16TB | Refurb $270 · Seagate Exos X16 16TB Renewed | $16.87 | $33.66-36.50 |
| 14TB | Refurb $240 · WD Ultrastar DC HC530 14TB SATA 4Kn NAS | $17.14 | $38.22-42.00 |
| 10TB | New $140 · MDD 10TB SAS 12G 7200RPM Enterprise | $14.00 | $47.99-59.33 |
| 12TB | New $230 · MDD 12TB SATA 6G 7200RPM NAS | $19.17 | $54.00-59.66 |
| 32TB | New $1,160 · Seagate SkyHawk AI 32TB Surveillance HAMR | $36.25 | $60.67-61.61 |
| 30TB | New $1,200 · Seagate Exos M 30TB SATA HAMR Enterprise | $40.00 | $68.61-69.64 |
The pattern the table usually reveals: skipping intermediate capacities and going straight to a larger drive often costs a little more up front but roughly halves how soon you will face the next upgrade, which is worth real money in a shortage where prices are elevated. For your exact drive count, current size, and RAID level, run the numbers in the upgrade planner.
Bigger drives mean longer rebuilds, and longer rebuilds mean longer windows in which your array is vulnerable. A 20TB drive can take the better part of a day or more to rebuild, and throughout that window a single-parity array has zero remaining redundancy. The statistical risk that matters is the unrecoverable read error, or URE: as you read every sector of every surviving drive to reconstruct the replacement, the chance of hitting a latent bad sector rises with the total data scanned — and at high capacities that total is enormous.
This is why RAID 6, with its second parity drive, becomes the sensible default at 16TB and above: it survives a URE or an outright second failure during a rebuild, exactly when RAID 5 would fail catastrophically. It is also why recording technology is non-negotiable — insist on CMR drives, because SMR drives rebuild so slowly (sometimes days) that they extend the danger window unacceptably. Plan the array with the RAID planner and verify recording tech on the CMR drive list before you buy.
Do not let the drives you pull sit in a drawer depreciating — in the current market they have done the opposite of depreciate. Sell the surplus while used prices are elevated; each old 8TB drive in our example is worth $51-68 right now, and that cash is exactly what makes the net cost of your upgrade so much lower than the sticker price. Keep one or two as cold spares so a future failure is a shelf-swap rather than a shortage-priced scramble, and consider repurposing one as an offsite backup target. Whatever you do, wipe each drive with a full-disk erase before it leaves your hands. The full per-capacity value ranges and where to sell are on the used drive value guide.
Yes, if you do it one drive at a time and let the array repair fully between each swap. In RAID 1, 5, 6, and 10, the array rebuilds the replaced drive's data from parity or mirror, so your data survives each swap. The absolute rule is: never remove a second drive before the previous rebuild has completed, and keep a verified backup in case a rebuild fails.
In standard RAID, yes. RAID 1, 5, 6, and 10 size the array to the smallest member, so usable capacity expands only after every drive has been replaced with the larger size and the final rebuild completes. Synology SHR and UnRAID are exceptions — they can deliver partial extra capacity after you replace the largest drives, without swapping the whole set.
Plan for 12 to 24 hours or more per drive at high capacities, and longer if the array is under load during the rebuild. A full four-bay upgrade therefore realistically spans several days to a week, because you must let each rebuild finish before starting the next. SMR drives rebuild dramatically slower — sometimes days — which is one reason to insist on CMR drives for RAID.
It depends on the layout. Standard RAID ignores the extra capacity of a larger drive until all members match, so mixing sizes wastes space there. Synology SHR and UnRAID are built to use mixed sizes — SHR pools uneven drives efficiently, and UnRAID simply requires the parity drive to be at least as large as the largest data drive. During an upgrade you will temporarily run mixed sizes, which is fine as an interim state.
Yes — buy every replacement drive up front. In the 2026 shortage, a mid-swap stock-out can strand you with a half-upgraded array running mixed sizes and no way to finish at the same model or price. Having all the drives on hand also lets you complete the sequence promptly and minimize the total time the array spends degraded across rebuilds.
In single-parity RAID 5 or SHR, a second failure during a rebuild means data loss — which is exactly why a verified backup before starting is non-negotiable. In RAID 6, which tolerates two simultaneous failures, you have a safety margin during rebuilds, which is why RAID 6 is strongly recommended once drives reach 16TB and rebuild windows grow long. If a new drive fails its rebuild, restore from backup and RMA the drive.
For a primary array holding your only copy of important data, buy new for the full manufacturer warranty and zero wear. For a secondary tier, a backup target, or any RAID 6 array with a hot spare, refurbished datacenter pulls deliver meaningfully better cost per terabyte at acceptable risk. Whichever you choose, insist on CMR recording — SMR drives are unacceptable for RAID because of their punishing rebuild times.