IBM 90Y8782 900GB SAS 10K G2HS HDD: Carrier and Compatibility Guide
Two IBM drives, same capacity class, same interface, same form factor, same speed. One seats in your chassis and one does not. The variable is four characters in the description that most people skim past: G2HS.
The IBM 90Y8782 is a 900GB SAS 6Gb/s 10K 2.5-inch drive supplied in a G2HS carrier, and the carrier generation decides whether it fits.
Carrier generations across the IBM and Lenovo transition
IBM used several hot-swap carrier designs across the System x lifespan, and Lenovo introduced another after acquiring the x86 business. They are not interchangeable, and the naming is not obvious from the outside.
G2HS is the second-generation hot-swap tray, used across a wide span of IBM System x servers. G3HS superseded it in the later IBM and early Lenovo era. Earlier IBM systems used other hot-swap designs again, and separately there is the simple-swap carrier type, which requires the server to be powered down and has no lever mechanism at all.
The differences are mechanical: latch geometry, connector alignment and the position of status indicator contacts. A drive in the wrong generation either will not slide home or will seat without registering correctly on the backplane.
The reliable check: pull a drive currently in the chassis and read the part number from its label. That number encodes drive, carrier and firmware together. Comparing trays visually works too, but the part number removes the guesswork.
90Y8782 specifications
| Part Number | 90Y8782 |
| Manufacturer | IBM |
| Capacity | 900GB |
| Interface | SAS 6Gb/s |
| Rotational Speed | 10,000 RPM |
| Form Factor | 2.5 inch SFF |
| Carrier Type | G2HS Hot-Swap |
| Drive Class | Enterprise Performance |
| Condition | Refurbished, function-tested |
| Warranty | 30-day TechSellerUSA warranty |
900GB at 10K: the balanced enterprise tier
10,000 RPM sits between the 15K performance tier and 7200 RPM nearline, and 900GB is a capacity point that reflects that positioning.
A 15K drive seeks faster, which matters for the most latency-sensitive transactional work, but 15K capacities are smaller and the drives run hotter and draw more power. Nearline drives hold far more but seek slowly enough to be unsuitable under a busy virtualisation host.
10K SAS at 900GB is the general-purpose enterprise answer: enough IOPS for virtualisation datastores, application servers and mid-sized databases, with enough capacity that you are not populating every bay just to reach a usable volume size.
In practice this is the tier most mixed-workload servers should be on, and it is why 10K SFF drives are the most commonly replaced part in the whole category.
SKU: 90Y8782 | Refurbished & function-tested | G2HS carrier | 30-day warranty
View Details & Request a Quote →ServeRAID firmware validation
IBM ServeRAID controllers check drive firmware, and a mechanically identical drive without IBM coding produces one of a few outcomes.
It may work while being flagged as unsupported in the management interface, which is cosmetically irritating and complicates any support conversation. It may work without reporting health telemetry correctly, so predictive failure warnings never reach you — which removes your chance to replace a drive before it fails rather than after. Or it may be refused during an array rebuild, which is the worst case because that is exactly the moment you are depending on it.
None of those outcomes is certain, and plenty of people run non-IBM drives without issue. For a lab or a non-critical file server, a generic drive with a scavenged tray is a reasonable saving. For anything in production with an array behind it, ordering by IBM part number removes the whole category of risk for a small price difference.
Sourcing for System x after the Lenovo transition
IBM sold its x86 server business to Lenovo, which split the support path and left a large installed base of System x servers with no straightforward parts channel.
Those servers are still entirely capable for virtualisation hosts, file serving, application servers and departmental workloads. Refurbished drives are usually the only realistic way to keep them populated, and holding a spare of each drive type in the chassis is standard practice.
The reason is simple: a failed drive in a degraded array is a race against the second failure, and a rebuild cannot start until a replacement is physically present. Every hour spent waiting on delivery is an hour spent exposed.
If your fleet spans the transition, our Lenovo range covers the later ThinkSystem equivalents, and the compatibility guide lists carrier types for both.
Ordering
We supply the 90Y8782 refurbished and function-tested with a 30-day warranty, single spares or in volume. If you are unsure whether your chassis takes G2HS or G3HS carriers, send us the part number from a drive currently fitted and we will confirm the match before you order. Browse the full IBM range or our complete enterprise hard drives catalogue.
Common questions
What does G2HS mean?
It is IBM’s second-generation hot-swap drive carrier, used across a wide span of System x servers. G3HS superseded it in the later IBM and early Lenovo era, and the two are not interchangeable — latch geometry, connector alignment and status indicator contacts all differ.
How do I check which carrier my chassis takes?
Pull a drive currently in the chassis and read the part number from its label. That number encodes drive, carrier and firmware together. Comparing trays visually works, but the part number removes the guesswork entirely.
What happens if I fit a non-IBM drive?
ServeRAID controllers check firmware. The drive may work but be flagged as unsupported, may work without reporting health telemetry so predictive failure warnings never arrive, or may be refused during an array rebuild. For production arrays, order by IBM part number.
When should I choose 10K over 15K or nearline?
10K is the general-purpose enterprise tier: enough IOPS for virtualisation datastores, application servers and mid-sized databases, with enough capacity that you are not filling every bay to reach a usable volume. Choose 15K for the most latency-sensitive work, nearline for backup and archive.
Are System x servers still worth maintaining?
For virtualisation hosts, file serving, application servers and departmental workloads, frequently yes. The hardware is capable; the constraint is parts. Refurbished drives are usually the only realistic supply, which makes holding a spare of each drive type standard practice.
Send us the part number from a fitted drive and we will confirm the carrier match before quoting.
