Seeing as I’m going to go on a holiday sometime next year, one thing I knew I would need is some portable external storage to keep all my experimental data and to backup the photos I will be taking. Not having purchased any external hard drives in a long time, it surprised me to see that the market had advanced to 4Tb 2.5″ portable drives. The unexpected part was that the largest capacity available also had the lowest cost per gigabyte, making it the best value for money. Traditionally, the largest drives tended to command a price premium, reflecting their desirability. As a result, I knew I would just have to have one.
Computer users of the early to mid 2000’s would probably not be fond of the Maxtor brand, as some of their models were somewhat unreliable. But since the Maxtor brand has been long acquired by Seagate and their factories integrated into the Seagate portfolio, the brand has been retained mainly as an alternative brand for external storage. As the Maxtor M3 Portable 4Tb drive was the most cost effective option, retailing at AU$209, I chose the product knowing in full confidence that internally, it would be very similar to the Seagate branded offering.
As every drive that enters service in my fleet undergoes “acceptance testing” for commissioning, I thought I might as well turn it into a review in the case anyone’s interested in just what’s inside and how it performs.
The unit comes boxed in a glossy cardboard box with a cut-out that shows the drive inside, protected by a plastic bubble. The cardboard box has a blue, black and white colour scheme, with the front showing some of the features including a three year limited warranty. The contents is listed on the rear, with some of the included “features” enabled by included software that is preloaded onto the drive.
The side spine has a label with the barcode and serial number of the drive, with the cut-out extending to the other side showing off the thickness of the drive. This makes it clear to any potential customers that this is a thick drive with an internal height of 15mm – this makes it less useful to those who might want to squeeze it into an older laptop or game console.
A close look at the underside makes it clear that the drive is made in China, by Seagate, as expected.
Inside the package, we have the drive itself featuring a multi-faceted plastic top cover design which is somewhat similar to the Seagate Expansion Portable drives. It also comes with a USB 3.0 super-speed cable (A to microB).
The drive has a connector on one side, which is slightly mis-aligned, and a window for the blue activity LED. The “middle” wrap-around plastic is glossy and attracts fingerprints, unlike the top and bottom.
Rubber feet are fitted to the bottom, but the profile is very thin, thus if placing on top of another drive with a curved surface, it has no grip. The drive serial number is also on the bottom. The drive weight is 241.16g as measured by my set of scales, making it fairly weighty but also similar to other drives of this size.
When the drive is plugged in, it was determined that the unit used a UASP capable chipset for better USB 3.0 performance with USB drivers featuring UASP capabilities (e.g. Windows 8 and above for most USB 3.0 controllers).
The hardware ID is VID 0BC2 PID 61B7. That VID is also used by other Seagate products.
The drive comes pre-formatted in NTFS and has software pre-loaded as promised.
The listing of files is below.
Because this is a more modern drive, released since the end of support for Windows XP, this drive doesn’t have that nasty 4k “sector translation” goof that other older drives did. As a result, it correctly reports 512 byte sector size with a 4096 byte physical sector size. This means that it’s likely that the drive can be “transplanted” from the enclosure to an internal SATA port with no compatibility issues.
Drive testing was performed on my Lenovo E431 (Intel Core i7-3630QM) laptop using onboard USB 3.0 ports and the latest Intel drivers with no other devices on the bus. The laptop is running the latest version of Windows 10 at this time (Anniversary Edition).
When I began testing, I found rates much slower than expected for write operations. This was because the setting for USB drives defaulted to quick removal, rather than better performance (which enables write caching).
Without write caching, you can expect write speeds at most half that of the expected speed. Enabling write caching requires diligent use of the “safe removal” features and acknowledgement that larger amounts of data loss can occur in the case of unexpected unplugging or power loss. This is an acceptable trade-off to me, and I advise using the “Better performance” option with “Enable write caching on the device” ticked to ensure the performance of the drive.
The test results that follow are all based on having the write caching configured on.
CrystalDiskInfo confirms that the drive contained within the enclosure is an ST4000LM016-1N2170, sometimes also sold as a Seagate/Samsung Momentus drive. This is a 5400RPM drive with 6Gb/s SATA interface and a whopping 128MB cache, consistent with that reported by the information above. The bare drive gets a 2-year warranty, so buying it in an external enclosure means it’s warranted for three. The weight is claimed as 200g for the bare drive, and it is revealed that typical power ranges from 1.1W (idle), 1.9W (read), to 2.1W (write).
Throughout the test regime, no SMART data failure indicators were recorded, indicating the drive is suitable for use.
The drive achieved an average of 103MB/s across the drive, ranging from 132.7MB/s down to 58.2MB/s. This is an improvement over some older drives I’ve used, but doesn’t seem spectacularly fast especially considering the density increase over the older drives.
Write performance was marginally slower, but otherwise much the same. The lumpy but almost continuous curve suggests this drive is using head-adaptive density techniques similar to those used by IBM/Hitachi in their drives to squeeze more capacity and reliability by avoiding “fixed zone bitrate formatting” constraints. It seems that the drive does do “unusual” things by responding that writes have completed as soon as they hit the buffer, resulting in an unrealistically low access time reported by the benchmark. The uncached shows a higher access time, probably due to USB latency, but still is flat across the drive suggesting the drive is acknowledging writes before the write has “hit the disk”.
Random access tests show that the drive’s read performance with random access is relatively limited, topping out at 95 IOPS over USB, whereas write operations seem more readily serviced, peaking at 1912 IOPS. I’m not too sure about this being actually consistent with long-term performance as caches can affect the accuracy of the result.
The extra tests seem to show the read cache buffer might not be working as effectively as possible, showing only speed gains for the first 1Mb and not much more thereafter. The write cache result shows an unusual dip, which was not expected. I wouldn’t pay too much attention to these results – they’re only provided for completeness.
I tend to pay more attention to CDM results when it comes to file performance. It seems UASP does make a good difference with read performance of small-blocks, with queueing allowing the performance to almost triple. Queued small-block write improvements were not as drastic. Sequential accesses was in line with what was seen with HDTune, with 512kB accesses showing significant read performance drop-off, and less so with write. Small block performance remains a challenge for mechanical drives – no surprise here.
ATTO is a bit of an industry standard, so I decided to include it anyway. It seems that 16kB accesses are well serviced, and even 8kB accesses do very well. 4kB accesses seem to struggle, and due to some odd behaviour, the 8192kB accesses are also somewhat slower. Perhaps, the drive format “alignment” was not quite correct to optimize performance despite the claim of Seagate drives having “smartalign” technology.
The drive was tested with H2testw mainly as a data integrity check. On the test platform, it successfully passed with no errors, but because of the size and speed of the drive (and potentially overheads with generating/checking the data), it took almost 15 hours to fill and 11 hours to verify. It was transported to my main desktop with an NEC Renesas USB 3.0 controller where it performed flawlessly and verified the full dataset with no faults.
As expected, the Maxtor M3 Portable Hard Drive is very much a Seagate product at its core, sporting a Seagate hard drive internally. The design of the casing is similar, and the internals seem to use a UASP capable SATA bridge which ensures better performance. It performed flawlessly in testing, although its speeds were relatively “average” when it comes to external hard drives in my experience with a few performance oddities. For the price of AU$209, it’s the best value per gigabyte, and it’s the largest portable bus-powered hard drive at the moment, which is a combination of attributes that are hard to beat. With a three year warranty, it seems highly recommendable as a large external storage solution.