Two hard drives with the same capacity, from the same manufacturer, sometimes sold at the same price, can still behave very differently once installed in a NAS or a RAID server. The reason comes down to a technical difference you won’t find on the label: how the data is physically written to the platters. This is the divide between CMR (conventional recording) and SMR (shingled recording). It’s a choice most buyers never think about, yet it has real consequences for performance, reliability, and — if a drive fails — for how likely it is that your data can be recovered.
CMR, the reference technology
CMR (Conventional Magnetic Recording) is the classic magnetic recording method, used since the early days of the modern hard drive. Each data track is written in isolation, without overlapping its neighbours. This means a track can be rewritten without touching adjacent tracks, which gives consistent, predictable write performance and good long-term behaviour. It’s the technology found in drives built for servers, demanding workstations, and professional RAID setups, where consistency matters more than squeezing out the maximum capacity at the lowest price.
Its limitation is purely economic: for the same platter size, CMR stores less data than newer technologies, at a higher manufacturing cost.
SMR: more capacity, at the cost of a trade-off
SMR (Shingled Magnetic Recording) was developed to meet the demand for ever-larger capacities without changing the physical size of the drive. The principle: partially overlap the write tracks, much like shingles on a roof, to fit more of them on the same surface. This tighter packing increases storage density, but it comes with a direct cost on writes: modifying one track often means also rewriting part of the neighbouring tracks it overlaps. SMR drives include internal logic (typically a CMR-style buffer zone, managed by firmware) to smooth this out, but once that buffer fills up under sustained write load, performance drops sharply.
SMR isn’t a lower-quality drive: it’s a drive optimised for a specific use case. It’s well suited to archiving, backups, or storing data that’s rarely rewritten, where capacity matters more than write speed. It’s much less suited to workloads involving frequent, sustained writes.
NAS and RAID: where the difference gets expensive
It’s precisely in NAS units and RAID arrays that the SMR trade-off causes the most trouble. A RAID rebuild after replacing a failed drive is exactly the kind of workload SMR handles worst: massive, continuous writing, for hours or even days.
The best-documented case remains Western Digital in 2020: several models in the WD Red 2–6 TB range, until then assumed to be standard NAS drives, turned out to be SMR drives, without this being clearly stated in the product documentation. Users who replaced a failed drive in their NAS reported abnormally long rebuild times, several days in some cases, with a higher risk of the rebuild failing outright. Western Digital later clarified its line-up (WD Red 8 TB and above remained CMR) and launched a dedicated range, WD Red Plus, entirely CMR, for demanding NAS/RAID use.
That precedent illustrates the rule worth remembering: for a NAS or a RAID array that will face a rebuild sooner or later, a CMR drive remains the safer choice, even at a slightly lower capacity or budget.
Why data recovery is trickier on an SMR drive
When a drive fails, the very structure of SMR complicates recovery work. On a CMR hard drive, each track is independent and can be imaged block by block in a fairly direct way. On an SMR drive, tracks overlap, and part of the mapping between logical data and its actual physical location is handled by the drive’s own firmware — an extra layer that can itself be damaged or become inconsistent when the drive fails.
Rebuilding usable data from a raw image of an SMR drive therefore requires methods and tools specific to this architecture, which not every lab handles to the same standard. It isn’t impossible, but it’s slower and more technical than a standard CMR recovery, which is why this type of case generally costs more.
How to tell whether your drive is CMR or SMR
This information is almost never displayed prominently on the packaging. A few reliable ways to check, before buying or on a drive already in service:
- The official lists published by manufacturers (Western Digital, Seagate, Toshiba), which state, model by model and part number by part number, which technology is used — check directly on the manufacturer’s site, as these lists change over time.
- The full datasheet for the exact model, which generally states explicitly “CMR” or “SMR” (or a manufacturer-specific term, such as Western Digital’s “DMSMR”).
- Absent any clear information, be wary of a significant price gap between two drives of the same capacity within the same range: it’s often a sign of a difference in technology.
For a NAS or RAID setup meant to last, it’s worth spending a few minutes checking this before buying, rather than discovering it the day a rebuild never seems to finish.
CMR or SMR isn’t inherently a good or a bad drive: it’s a technical choice that needs to match the actual use case. The problem arises when that choice is made blindly, on a drive destined for intensive writes and RAID rebuilds. And if the failure has already happened, the nature of the drive directly changes the approach needed to give your data the best chance of being recovered.