The short answer: In April 2020 Western Digital published a list of its SMR models, and it holds nine model numbers, nothing more. WD40EFRX is not on that list, so it is a CMR drive. WD40EFAX is on it, so it is SMR. Those two letters in the middle are the entire difference between a drive that belongs in a RAID and a drive that does not. Seagate and Toshiba have their own official tables — you will find them side by side below, arranged so that you can look a drive up by the model number on its label.
This article does not explain how SMR works. That is covered in the overview of CMR and SMR technology. What matters here is the single question people bring to us most often: I have a specific drive on the bench with a specific model number — is it SMR or not?
The official WD list: nine models and one sentence
After the pressure that built up in 2020 around unlabelled SMR drives in the Red line, Western Digital released a single page with a list on it. The slide is called „WD Internal Client SMR HDDs", it is dated 22 April 2020, and this is what it contains:
| Line | Capacity | Model number |
|---|---|---|
| WD Red 3.5” | 2 TB | WD20EFAX |
| WD Red 3.5” | 3 TB | WD30EFAX |
| WD Red 3.5” | 4 TB | WD40EFAX |
| WD Red 3.5” | 6 TB | WD60EFAX |
| WD Blue 3.5” | 2 TB | WD20EZAZ |
| WD Blue 3.5” | 6 TB | WD60EZAZ |
| WD Blue 2.5” | 1 TB | WD10SPZX |
| WD Blue 2.5” | 2 TB | WD20SPZX |
| WD Black 2.5” | 1 TB | WD10SPSX |
Underneath the table sits a sentence that matters more for a model-number search than the table itself: „All other capacity points use CMR recording technology." In other words, every other capacity uses CMR. Source: WD Internal Client SMR HDDs, PDF.
Two answers people search for most often follow from that:
- WD40EFRX (the older WD Red 4 TB) is not on the list → it is CMR.
- WD10EZEX (WD Blue 1 TB, 7,200 rpm) is not on the list → it is CMR.
Confusing WD40EFRX with WD40EFAX is the most common mistake in this whole subject. They differ by two letters, they were sold in the same box under the same line name, and one belongs in an array while the other does not. Before you order „the same drive that was in there", read the full model number off the label, not the name of the line.
Lookup table by manufacturer and line
| Manufacturer and line | Technology | Source |
|---|---|---|
| WD Red 3.5”: WD20EFAX, WD30EFAX, WD40EFAX, WD60EFAX | SMR | WD slide |
| WD Red 3.5”, other capacities (e.g. WD40EFRX, WD60EFRX) | CMR | same source, note below the table |
| WD Red Plus — the whole line, 2 to 12 TB | CMR | WD Red Plus product brief |
| WD Red Pro | CMR | WD slide, note below the table |
| WD Blue 3.5”: WD20EZAZ, WD60EZAZ · WD Blue 2.5”: WD10SPZX, WD20SPZX | SMR | same source |
| WD Blue, other capacities (e.g. WD10EZEX) | CMR | same source, note below the table |
| WD Black 2.5”: WD10SPSX | SMR | same source |
| Seagate BarraCuda 3.5”: 4 TB and 8 TB | SMR | Seagate CMR/SMR list |
| Seagate BarraCuda 3.5”: 12, 16, 20 and 24 TB | CMR | same source |
| Seagate BarraCuda 2.5”: 2 TB and 4 TB · FireCuda 2.5”: 1 TB and 2 TB | SMR | same source |
| Seagate Archive 3.5”: 8 TB | SMR | same source |
| Seagate IronWolf 2–16 TB · IronWolf Pro 2–24 TB | CMR | same source |
| Seagate SkyHawk 1–8 TB · SkyHawk AI 8–24 TB | CMR | same source |
| Seagate Exos 4–24 TB (and Mozaic HAMR 28/30/32 TB) | CMR | same source |
| Seagate Exos 32, 36 and 44 TB | SMR | same source |
| Toshiba P300, DT02 and DT02-V: 4 TB and 6 TB | SMR | Toshiba announcement |
| Toshiba L200 and MQ04 (2.5”): 1 TB and 2 TB | SMR | same source |
| Toshiba N300 | no SMR | same source |
Only what the manufacturers have published themselves belongs in this table. If a model appears on none of those three lists, look it up in the datasheet for that exact model number — not by the name of the line.
Seagate has to be looked up differently from WD
A difference worth noting: Western Digital published a list of model numbers, whereas Seagate keeps its table by line and capacity. With Seagate you therefore will not find your specific SKU in the official list — you will find the line and the capacity.
For the SMR BarraCuda 3.5” capacities, the models commonly sold are ST4000DM004 (4 TB) and ST8000DM004 (8 TB), and ST2000DM008 for the two-terabyte variant. Take that as a pointer to where to start: read the line and the capacity off the label and compare them against the official Seagate table. Manufacturers rotate SKUs within a single line and capacity over time, and retail listings routinely garble model numbers.
It is also worth mentioning that at Seagate, SMR has recently returned where you would not expect it — in the Exos line, the 32, 36 and 44 TB capacities are listed as SMR, while 4 to 24 TB are CMR. So „enterprise line = CMR guaranteed" no longer holds.
WD Red Plus: the line you do not have to ask about
After 2020 Western Digital introduced the Red Plus line as an explicitly CMR variant (background on the whole affair at Tom’s Hardware). In the Red Plus product brief, all ten models today carry the value CMR on the „Recording technology" line: WD120EFGX and WD120EFBX (12 TB), WD100EFGX and WD101EFBX (10 TB), WD80EFPX and WD80EFZZ (8 TB), WD60EFPX and WD60EFZX (6 TB), WD40EFPX (4 TB) and WD20EFPX (2 TB).
Parameters shared by the whole line according to the same document: a declared workload of 180 TB per year, MTBF of 1,000,000 hours, a 3-year warranty, 600,000 load/unload cycles and non-recoverable read errors below 1 in 10^14 bits read. Which of those models makes sense for your number of bays is dealt with in the comparison of WD Red, IronWolf and Toshiba N300.
Why it matters in an array
This is not about speed during everyday work. It is about the moment when the array recomputes the contents of a replaced drive — the rebuild, or the resilver in ZFS. That is when the drive is hit by a long sequential write of exactly the kind that gives an SMR drive the most trouble.
A documented case is OpenZFS issue #10214, titled „WD WDx0EFAX drive unable to resilver". The user who reported it describes the state of a resilver on a WD40EFAX drive in these words: „947G resilvered, 27.61% done, 1 days 13:03:18 to go", adding that IOPS had dropped to zero while the write error count kept climbing — the operation had therefore stalled, even though the kernel went on reporting an estimate of the time remaining. Source: openzfs/zfs#10214 on GitHub.
You will not find specific numbers along the lines of „an SMR rebuild takes X hours and a CMR one Y hours" in this article. We were unable to track down a measurement that could be backed by a primary source, and an estimate would be useless here — what decides it is the capacity of the array, how full it is, the type of workload and whether anyone is working through the array during the rebuild.
Why the rebuild is precisely the most dangerous moment, and what goes on inside the array, is examined in Why a rebuild kills RAID arrays.
How to find out what is in your NAS right now
The model number is on the drive label, but you do not have to take the array apart to get at it. Reading it out of the system is enough:
- Linux, Synology or QNAP over SSH:
smartctl -i /dev/sdaprints aDevice Modelline with the full model number. For all drives at once,for d in /dev/sd?; do smartctl -i $d | grep -E "Device Model|Serial"; done. - Windows: CrystalDiskInfo shows the model string in the header of each drive.
- DSM and QTS in the web interface: Storage Manager → the drive list, the model column.
Compare the model string against the table above. If it is not there, two pointers from the manufacturer apply: Western Digital states in its support pages that „Legacy HGST and WD hard drives built before 2018 use CMR" — the manufacturing date is on the label — and otherwise recommends checking the datasheet of the specific model (WD Support, Steps to Determine if an Internal Drive uses CMR or SMR Technology).
What you cannot tell: the recording technology from a SMART readout. No attribute reports it. Nor is speed in a short test a pointer — an SMR drive takes data into a fast area and only slows down once that area fills up.
When the array already has SMR in it
Let us assume you have just read WD40EFAX off three of the four bays. Proceed in this order:
- Do not start anything extra. No scrub, no benchmark, no bulk copying. An array that is still working should be left to work the way it has been working so far.
- Check your backup. RAID is not a backup. If the only copy of the data is in this array, the work starts here and everything else waits.
- Do not replace all the drives at once. Every replacement means a rebuild, and the rebuild is the riskiest operation in the life of an array.
- Add CMR. As soon as you replace any drive, put a CMR unit from the table above in its place. The procedure for a gradual move to different drives without losing data is described in expanding a NAS with larger drives.
- Watch out for drives on their way out. Which SMART values are meaningful and which are not is summarised in How to tell that an HDD or SSD is dying.
If the array is already reporting a degraded state, do not start by replacing a drive — first find out why the drive dropped out of it. For Synology, the route to that is the procedure for Storage Pool Degraded and Volume Crashed.
Frequently asked questions
Is WD40EFRX SMR or CMR?
CMR. It is not on Western Digital’s official list of SMR models, and the note below the table says that all other capacities use CMR. Watch out for WD40EFAX, which is on the list and differs by two letters.
Is WD10EZEX SMR?
No. The WD Blue 1 TB under the designation WD10EZEX is not on WD’s list of SMR models, so it is a CMR drive. From the WD Blue 3.5” line, only WD20EZAZ and WD60EZAZ are SMR.
Can I tell SMR from the SMART values?
No, SMART does not report the recording technology. Read the model number (smartctl -i, CrystalDiskInfo) and compare it against the manufacturer’s table.
Can I at least use an SMR drive for backups?
For an archive copy that is written in large chunks and rarely, an SMR drive behaves well. What is problematic is a deployment involving a long sequential write into a full drive — that is, an array rebuild.
Is Seagate IronWolf always CMR?
According to Seagate’s official table, IronWolf 2–16 TB and IronWolf Pro 2–24 TB are listed as CMR. Always verify the capacity you are buying, though — Seagate keeps its list by line and capacity, not by model number.
When the array has failed to come up, the rebuild has stalled or the volume is reporting an error, stop making further repair attempts — every additional write reduces the chance of a complete recovery. We can read the drives outside the NAS and put the data back together: data recovery from a RAID array and data recovery from a Synology NAS. If, on the other hand, you are only just building the array and want the drive line-up checked, write to miroslav.jaros@ithope.cz with the model list from smartctl -i — going through a list of drives against the official tables is a matter of minutes.
Where to find us
ITHOPE is a company based in Brno, Czechia: we do data recovery from disks, RAID and NAS, PC and laptop service and IT for business, all in our own lab at Viniční 3587/216, 615 00 Brno.
Phone +420 775 556 063 (24/7 line). You can bring the device in person, or send it by courier (Zásilkovna, PPL).