by Andrey Kuznetcov
04/11/2006 | 07:05 PM
While the size of traditional 3.5” hard disk drives increases, the external solutions based on them also keep adding up their storage capacity. Among the Maxtor hard drives the leadership in terms of storage capacity belongs to Maxtor OneTouch III Turbo with the capacity of 1TB. This value is actually a sort of a psychological threshold, when the storage capacity is measured not by the common gigabytes, but by principally new units.
If you are looking for the most optimal external storage solution that could accommodate a lot of data, then this review might be very helpful, as it will reveal all the details about the new Maxtor solution.
Before we dig into the details about the new representative of external Maxtor hard disk drives, I would like to mention that this HDD family includes not only an unprecedentedly big 1TB solution but also a 600GB solution.
The exterior design of the OneTouch III Turbo Edition is very similar to other external OneTouch III drives. The only difference between them is the thickness of the device: our hero is considerably thicker. The reasons are evident: there are two 3.5” hard disk drives inside the HDD case instead of one. We are dealing with a square box with ventilation slits on the side panels. The front panel of the casing features Matrox’ brand name OneTouch button, that gave name to the entire product line. By pressing this button you launch automatic data archiving. Besides this major function, the button also plays an additional informational role. The LED status indicator is built into this button. There are two FireWire 800 ports, a FireWire 400 port, a USB 2.0 port? A power supply connector and a power switch on the back side of the case. The back panel is also provided with numerous ventilation slits.
The two hard disk drives used in the external Maxtor OneTouch III Turbo Edition solution can be configured to build two types of RAID arrays. In the first case we can use only RAID 0 (this is our today’s case) and in the second case the user can choose between RAID 0 and RAID 1. The HDDs’ spindle rotation speed equals 7,200rpm, and the buffer is 16MB big. Their average seek time doesn’t exceed 9ms. The external hard drive supports three interfaces at once (thanks to the Oxford 924 chip). The first interface is USB 2.0 with the 480Mbit/s theoretical bandwidth that can guarantee 34MB/s in practice. The second interface is FireWire 400 with 400Mbit/s theoretical bandwidth and 43MB/s practical data transfer rate. And the third interface is FireWire 800 that can provide 91MB/s data transfer rate at 800Mbit/s. The drive can work in the environmental temperatures varying between 5 and 35o C. Its dimensions are 136x98x217mm and it weighs 2.6kg.
The external hard drive is shipped with an external power supply unit with a power cable, a quick installation user’s guide, two FireWire cables, a USB cable, a CD disk with the EMC Retrospect Express HDD backup software and user’s manual.
The approximate retail price of the device is around $800.
Since the hard disk drive we are going to review today supports all three interfaces at the same time, we decided that it would be interesting to see what the actual performance of the device is depending on the interface it is connected through.
We tested the external drive in the following benchmarks:
When the hard drive worked via USB 2.0 and FireWire 400 interfaces the testbed was configured as follows:
When the hard drive was connected via the FireWire 800 interface the testbed configuration looked as follows:
The drive worked with the FireWire 800 interface via the PCI IEEE 1394b card (Texas Ins., TSB82AA2 chip). The FireWire 400 interface support was provided by the VIA VT6307 chip integrated onto the mainboard. The tests with USB 2.0 interface were conducted by connecting the drive to the USB 2.0 port on the mainboard. During our test session the hard drive was formatted for FAT32 and NTFS file systems with the cluster of default size.
I would like to draw your attention right away to the fact that we used two platforms during this test session. These platforms differed from one another not only on the hardware level, but also in the software. The latter is actually more important for us. Since Windows 2000 doesn’t have the option for enabling maximum performance for the external storage solution that we have in Windows XP. Of course, this factor will have its influence on the benchmark results, which we will point out later during the tests discussion.
During linear reading tests, Intel IOMeter sends a stream of read/write requests with a request queue depth of 4 to the hard drives. The data block size changes every minute. As a result, we can see the dependence of the drive’s sequential read/write speed on the size of the processed data block. Of course, we would also like to reveal the influence imposed by different types of the interfaces involved.
The hard drive performance best of all when it uses FireWire 800 interface. In this case the linear read speed is maximal for the data blocks of any size. FireWire 400 interface can only compete with FireWire 800 interface on small data blocks. After that it falls considerably behind. When we use USB 2.0, the potential performance of the storage solution gets limited most of all. In this case the linear read speed of the drive appears the lowest.
Nothing new can be found on the linear write speed diagram. FireWire 800 is an indisputable winner here and the drive performs best of all in this case. FireWire 400 interface can guarantee the same efficiency only with 512 Byte data blocks. After that linear write speed starts dropping. USB 2.0 interface is unable to reveal the full potential of the drive. The write speed is considerably lower in case this interface is involved.
This pattern sends a stream of requests to read and write 8KB random-address data blocks. During the test, the percentage of read and write requests in the stream changes.
The obtained results allowed us to build three diagrams for more illustrative analysis. The first diagram demonstrates the situation when the requests queue depth is minimal throughout the test. All three curves are extremely close to one another. We cannot claim that any of the interfaces is better than the others. We can observe efficiency parity here.
With the queue depth equal to 16 requests USB 2.0 interface boasts a slight advantage. In this case the hard drive shows better results during processing of requests with high reads share.
In case of the maximum queue depth, FireWire 400 and FireWire 800 interfaces show the best results. They appeared slightly more efficient than USB 2.0.
During the workstation emulation tests, the hard drive workload was limited to only 32 requests queue. The results were taken for the entire hard disk drive storage capacity as well as for the 32GB address space.
In order to give you a better idea of the interfaces efficiency we built the performance rating diagram with the results calculated by the following formula:
Performance Rating = Total I/O (queue=1)/1 + Total I/O (queue=2)/2 + Total I/O (queue=4)/4 + Total I/O (queue=8)/8 + Total I/O (queue=16)/16 + Total I/O (queue=32)/32
The first diagram shows the hard drive’s performance when we use up the entire storage capacity of the drive. The performance difference is not that dramatic in all three cases. The advantage of FireWire 800 interface over FireWire 400 interface is nominal. The use of USB 2.0 interface pushes our hero farther behind.
The second case we took into consideration here is when the storage capacity in question is limited to 32GB. The results remain practically the same, but the overall performance increases, which is quite logical. The performance difference between different interfaces is again quite small, but the maximum results are again obtained with FireWire 800, and the minimum results – with USB 2.0.
File- and WebServer patterns allowed us to emulate the typical file- and webservers performance.
Two diagrams below show the integral performance ratings of the drive that were calculated as the average requests processing speed for all types of workload.
The emulation of the file-server workload showed that the performance results obtained for three different interfaces hardly vary. The hard drive runs just a little bit faster when it is connected via the FireWire 800 interface compared with what we see when it runs via FireWire 400. The performance drops a little more when we switch to USB 2.0.
When we emulate the web-server operation the situation is very similar again. FireWire 800 interface is the fastest, while FireWire 400 is falling just a little bit behind. The lowest results were obtained with USB 2.0 interface.
During multi-threading tests we emulated the situations when there are up to 4 users working with the drive at the same time. The drive would perform both: read and write requests with 64KB data blocks within 30 seconds. The number of requests was gradually changed from 1 to 8.
We decided to refrain from building diagrams for all situations we emulated. Instead we would like to specifically dwell on those cases when the request queue depth was minimal, as this is most typical for real life.
As you can see, no matter how many users there are, the maximum performance during reading is achieved with FireWire 800 interface. The drive works much faster in this case than with the other two interfaces. Of course, FireWire 400 is more efficient for multi-threaded reading than USB 2.0.
Nothing new can be seen on the multi-threaded writing diagram. FireWire 800 interface is undoubtedly the best choice here. As for FireWire 400, it once again proved to be more efficient than USB 2.0.
when we tried to test a hard drive of this size with an old benchmarking tool like Winbench 99, we faced some difficulties connected with the benchmark’s inability to correctly display the entire storage capacity of the drive. Therefore, we limited the size of the test space to 32GB.
Let’s start discussing the results of Winbench 99 with the surface read speed graphs.
The diagrams show very well the actual peak bandwidths for each interface. I can admit that in this case the results are very close to the claimed specifications. By taking a closer look at the graphs above we can say that with USB 2.0 interface the peak value reaches 33MB/s, with FireWire 400 and FireWire 800 – 42MB/s and 90MB/s respectively.
Now let’s take a look at the hard drive performance on the first 32GB of storage space. Let’s start with the results obtained for FAT32 file system.
The diagram above shows very well the advantages of this storage solution when it is hooked up via FireWire 800 interface. The same is true about the advantage of FireWire 400 over USB 2.0: the results speak for themselves.
In NTFS file system the situation is slightly different. The advantages of FireWire 800 interface over FireWire 400 interface are almost negligible. Of course, the use of slower file system explains the overall lower results. However, the use of two different operating systems could also tell on different performance results for the used interfaces.
The last diagram shows the surface read speed. Of course, the best results are achieved with FireWire 800 interface. The results obtained with FireWire 400 are noticeably lower. USB 2.0 is the least efficient of all.
As always, we will wind up our tests with the results from FC-Test utility, which provides the most realistic performance results and efficiency estimates for each given interface. This program is based on measuring the time it takes the drive to create (write), read and copy certain file sets that differ from one another by the size and number of files. After that we calculate the hard drive performance basing on the time measurements taken.
As you probably already now, Windows and Programs file sets include a large number of smaller files, while other three patterns contain a limited number of larger files. During file copy operations the hard drive is split into to logical partitions 32GB each. The file sets are either copied within the same partition or from one partition to another.
First we will check out the results obtained for FAT32 file system.
The first diagram shows the results obtained during file writing (creation). The picture that we saw is exactly what our theoretical expectations from this test were. Our hero creates (writes) files fastest of all when it is connected via the FireWire 800 interface. The results are much lower when we switch to FireWire 400 or USB 2.0 interface. We weren’t at all surprised that the actual efficiency of the FireWire 400 interface was again higher than that of the USB 2.0.
The next diagram shows the file read speeds. The situation is generally identical to what we have just seen in the previous case. Once again the maximum speed is achieved when we connect the hard drive via the FireWire 800 interface. The results are much lower in case of FireWire 400, and even lower in case of USB 2.0.
The third diagram shows the performance of our hard drive connected via three different interfaces during file copying within the same partition. It is quite natural that all the results are generally lower than in the previous two cases. However, the picture has hardly changed: FireWire 800 connection is the most efficient. When the HDD works via FireWire 400 interface its results are much lower, but the numbers are still better than in case of USB 2.0.
Nothing really new, when we get to copying files from one partition to another. FireWire 800 interface is again indisputably better than FireWire 400 and USB 2.0. And again the drive works faster when connected via FireWire 400 than in case of USB 2.0.
Now we should pay some attention to the results obtained in NTFS file system.
On the diagram with the file writing speed FireWire 800 interface proved highly efficient in all five patterns. The lowest results were obtained when we connected our hard drive via the USB 2.0 interface. It is remarkable that the performance difference between the two FireWire interfaces is relatively low when we deal with writing a lot of smaller files: Programs and Windows patterns. It is probably the linear write speed that we have already measured earlier, which doesn’t differ that much when we work with smaller data blocks.
The situation during file reading is very similar to the previous case. FireWire 800 interface appeared the best-performing here. Once again it is not that much ahead of the FireWire 400 interface when we work with Programs and Windows patterns. It is probably again the insignificant linear read speed difference between these two interfaces that determines results like that for smaller data blocks. USB 2.0 interface is the slowest of all once again.
When we copy files within the same partition in patterns that deal with large number of smaller files, FireWire 800 interface manages to lose to FireWire 400, and in one pattern it is even slower than USB 2.0. Nevertheless, USB 2.0 is the least efficient of all three interfaces.
The last diagram shows the performance of our hard drive connected via different interfaces when we copy files from one partition to another. If we look at the two patterns that include a lot of smaller files we will see that FireWire 800 yielded not only to FireWire 800, but also to USB 2.0.
We tested the Random Read speed in Programs pattern that consists of all those files you would usually find on any PC. The idea behind this test is to see how fast the fragmented hard disk drives could be. During the test session, the program performed random reads of file successions made up of ordered 1-10 files. The only randomness limitation was the peak set for every 100 files on the next read.
This diagram shows that FireWire 400 interface appeared the most efficient in all other cases except the single file reading. Yes, this particular interface and not its newer counterpart – FireWire 800 – won the test. The results were significantly less impressive when we used USB 2.0 interface. As I have already supposed above, the less impressive performance of the FireWire 800 interface could have been caused by the different operating system we used on this platform.
The external Maxtor OneTouch III Turbo Edition hard drive with 1TB storage capacity will undoubtedly find its niche in the market. There are hardly any external solutions out there that can offer that much storage space today. The portability of this drive and the support of three interfaces make it very suitable for archiving and transportation of large data chunks when the configuration of the system is unknown. High speed of this solution is also an advantage. FireWire 800 interface and RAID 0 support allow transferring up to 1TB of data within an acceptable period of time. Although we have just shown in our test session that FireWire 800 interface wasn’t faster than FireWire 400 in some cases, however, it may also result from the different operating systems we used. The widely spread USB 2.0 interface with higher theoretical bandwidth than FireWire 400 once again turned out less efficient, however it will still be needed if there is no FireWire connector in the system.
Since Maxtor OneTouch III Turbo Edition family offers devices that support RAID 0 or RAID 1, you can always choose what is best for your needs. Of course, the price of solutions like that is quite high and hence not so many users will go for an external piece of storage that expensive. So taking into account the specifics of this product I would assume that it is going to become more popular in the corporate market rather than in the home solutions market.