P53552-B21 HPE 20TB Drive: Rebuild Time and RAID Choice
The P53552-B21 is an HPE-coded 20TB hard drive for ProLiant servers. At this capacity there is one consideration that outweighs everything else, and it is not performance.
Rebuild time is the design constraint
The point to understand before ordering.
When a drive fails in a RAID group, the array rebuilds onto a replacement by reading every remaining drive in full. At 20TB that read takes a very long time — frequently more than a day, sometimes considerably more depending on array load.
Three consequences follow, and they change the design.
The array is degraded throughout. No redundancy, and a second failure during that window loses the group.
Every surviving drive is under sustained load for the whole rebuild — which is precisely the stress that causes second failures.
Drives installed together age together. They share a manufacturing batch, operating conditions and hours run, so when one reaches wear-out the others are at the same point. Our lifespan guide covers this correlation.
Put together: RAID 5 is not appropriate at this capacity. Single parity means one failure during a day-long rebuild loses everything. Use RAID 6 or equivalent double-parity, which tolerates a second failure during the rebuild — our RAID guide covers the arithmetic.
This is not caution. It is the standard position on high-capacity nearline storage.
What a 20TB drive is for
Drives at this capacity are nearline — built for capacity at 24/7 duty rather than for low latency.
They suit backup targets, archives, media stores, surveillance retention, file shares and bulk capacity tiers. Sequential throughput is good; random access is what a mechanical drive is inherently poor at.
They are the wrong choice for transactional workloads — databases, virtualisation datastores, busy application servers. Those need flash or at minimum faster mechanical drives, and putting a VM host on nearline storage is one of the more common causes of a system that feels slow for no obvious reason.
Confirm the interface, rotational speed and workload rating against the drive documentation for your specific part before ordering.
Fewer large drives is not the same as more small ones
A trade-off worth thinking through at this capacity.
Twenty terabytes in one drive uses one bay. The same capacity in smaller drives uses several — more power, more heat, more bays consumed.
But capacity is not the only thing you buy with a drive. More drives means more heads working, which means more requests served per second. A tier built from fewer large drives has less performance than the same capacity spread across more.
And more capacity per drive means longer rebuilds, per above.
So the honest position: large drives are right where capacity is the requirement and access is sequential. Where the tier needs to serve requests, spindle count matters more than capacity per drive.
The compatibility checks
Three, and they are the same ones that govern any ProLiant drive.
Carrier generation must match your chassis. A drive with correct capacity and interface will not seat if the Smart Carrier generation is wrong. It is encoded in the part number rather than the description — our part numbers guide covers this.
The chassis must take 3.5 inch drives. High-capacity nearline drives are large-form-factor, and a small-form-factor chassis cannot take them at all — our backplanes guide covers why bay size is a chassis decision.
The controller must support the capacity. Older Smart Array controllers have limits, and a drive above them may be unusable or may present incorrectly. Check before ordering — our controller guide covers identifying what is fitted.
HPE Smart Array controllers also validate drive firmware, which is the argument for an HPE-coded drive over a bare equivalent where one is fitted. Our OEM guide covers the distinction.
Practical points on high-capacity drives
Three worth knowing.
Advertised capacity and reported capacity differ. A drive sold as 20TB reports less in operating systems that count in binary units — this is a units difference rather than missing capacity, and our capacity guide covers it. Plan against the reported figure.
Physical handling matters. Large drives are heavy and shock-sensitive, particularly when running. Our handling guide covers the discipline.
Hold a spare. On a tier this size, a spare on the shelf turns a day-long degraded window into a shorter one. Our spares guide covers weighting a holding by consequence.
Before you order
Your ProLiant model and generation, and whether it takes 3.5 inch drives. What controller is fitted and its capacity limit. What RAID level you intend — and if the answer is RAID 5, reconsider. Whether you need carriers. And whether the workload is genuinely sequential, because if it is not, this is the wrong drive class.
Common questions
Can I use RAID 5 with 20TB drives?
You should not. A rebuild at this capacity takes a very long time, during which the array has no redundancy and every surviving drive is under sustained load — which is exactly when second failures happen. Use RAID 6 or equivalent double parity.
Is one large drive better than several smaller ones?
For capacity and bay efficiency, yes. For performance, no — more drives means more heads and more requests served per second, and larger drives mean longer rebuilds. Large drives suit capacity tiers with sequential access.
Will my controller support a drive this large?
Check before ordering. Older Smart Array controllers have capacity limits, and a drive above them may be unusable or present incorrectly. The chassis must also take 3.5 inch drives, since high-capacity nearline drives are large-form-factor.
Can I use this for my virtualisation host?
It is the wrong drive class for that. Nearline drives are built for capacity at sequential access, and a VM host issues heavily random requests. Putting a host on nearline storage is a common cause of a system that feels slow for no obvious reason.
Why does it show less than 20TB?
A units difference rather than missing capacity — drives are sold in decimal terabytes and many operating systems report in binary units. Plan your capacity against the reported figure rather than the advertised one.
Send us your ProLiant model, generation and controller, and we will confirm carrier, capacity support and whether the drive class fits your workload.
