by Andrey Kuznetcov
06/08/2004 | 02:53 PM
With huge masses of information around us, the requirements to storage media have grown considerably. When transferring data from one computer to another, not on a network, you often find yourself copying all files onto your hard disk and detaching it and carrying it to the other machine. Hard disk drives are oftentimes preferable storage media due to two their properties: high capacities and high data-transfer rates. The capacity bar is 400GB for today’s devices, but this doesn’t seem too much when you take to video editing in real time or some other resource-consuming activities.
So what does the user have to do if he/she needs to store and transport extra-large amounts of binary digits? Our today’s review is going to answer this question and help you choose a device optimal from several points of view, including price, performance, easy connection and transportation. A size of 300GB is going to be our starting point: we took devices from Maxtor Corporation for our tests as they meet our requirements and, until recently, have had no competitors as concerns the maximum capacity. So how may these 300 gigabytes of storage space be available to you?
First, you can use an ordinary IDE-interfaced hard disk drive, the 5A300J0 model number. This solution means the lowest cost, highest data-transfer rates and minimal dimensions of the transported device. The disadvantages are the inconvenience of the process of plugging/unplugging the drive (you have to open the system case up, if you’ve got no HDD rack) and the risk of damaging the drive during transportation.
Second, quite contrary to first, you can buy an originally and truly external device, like Maxtor’s OneTouch drive. The pros of this solution include easy connection across FireWire or USB 2.0 interfaces, protection of the device against external physical impacts, and a case design to tickle your refined aesthetic feeling. The cons of this solution are high cost and lower performance.
The third variant is an in-between one. You can use the same 5A300J0 drive in a separately-obtained case with support of the modern interfaces, USB 2.0 and FireWire. This variant seems to be close to Solution 2, but it also seems to save you some money (if you go for a not very expensive container).
The manufacturer intends this device for “near-line and other low-I/O secondary storage applications”. Its capacity is 300GB; the spindle rotation speed equals 5400rpm; the cache buffer is 2MB big; the drive’s average access time is below 10msec. This MaXLine supports the UltraATA/133 interface. The device can withstand a maximum operational shock of 60G for 2msec, and a non-operational shock of 300G for 2msec.
The average retail price of this drive is $280.
The external hard disk drive of the OneTouch family is based around the same MaXLine II 5A300J0 drive and thus shares all its technical characteristics, enumerated above. The difference is in the interface – the external model connects through either FireWire or USB 2.0 (USB 1.1), which limits the data-transfer speed somewhat. The peak sustained transfer rate is 34MB/s for USB 2.0 and 41MB/s for FireWire, according to the manufacturer, – these numbers are of course lower than the theoretical ceilings for the interfaces.
The accessories include a USB cable, a FireWire cable, a power adapter, a stand for holding the drive upright, a user manual and two CDs with software.
The manufacturer prices this device at $341.
Our third participant is a special container, intended to accommodate a 3.5” HDD. You can attach this thing via FireWire or a USB 2.0 interface. The manufacturer doesn’t mention any performance numbers, though. The case allows placing the drive vertically. The dimensions of the STLab case are 115x35x220mm. You use it with an external power adapter, which you receive along with USB and FireWire cables, a software CD and a technical description.
The average retail price of the case is $67.
Today we are going to check out performance of hard disk drives in five different configurations. First, we take a Maxtor MaXLine II 5A300J0 HDD and connect it to its native Ultra ATA interface. Then we will seal it into the STLab rack and test it attaching to two interfaces (FireWire and USB). The external Maxtor OneTouch drive is also tested with those two interfaces.
For our tests we used the following software:
The testbed was configured as follows:
The external devices were attached to the mainboard’s integrated USB 2.0 interface (realized through the ICH5) and to a FireWire port on a PCI add-on card with the VIA VT6306 chip. We carried out our tests using the default drivers of the operating system. The drives were formatted in FAT32 and NTFS as one partition with the default cluster size. In some cases, specially mentioned below, we used logical volumes of 32GB capacities, formatted in FAT32 and NTFS with the same default cluster size. The Maxtor drives were “written through” before the tests to avoid their forced write verification, which would otherwise influence the results.
We examined two situations with WinBench: 1) using the full capacity of the drive and 2) using 32GB of their capacity.
Regrettably, we suffered casualties in this test as we couldn’t get the numbers for the Maxtor 5A300J0 installed into the STLab case and attached via FireWire. The controller of this case must have been the reason for that.
Let’s first examine the numbers we got in FAT32. Using the entire capacity of the media, we have a natural outcome: the Maxtor 5A300J0 connected via the Ultra ATA interface wins the test. It is much faster than the drives that work in other tested configurations, both in High-End Disk WinMark, which we consider a priority score, and in Business Disk WinMark. The external Maxtor OneTouch drive attached via FireWire is the second in this test – it is a bit faster than itself attached via USB 2.0. The winner, the Maxtor 5A300J0, becomes the slowest device when you install it into a STLab container and connect through USB 2.0.
You may note that the access time of the external Maxtor OneTouch drive is much higher than with the Maxtor 5A300J0. This 5-6msec difference is probably because of the fact that the drive in the Maxtor B14D300 works in the low-noise mode (it moves it heads with some latency to reduce the acoustic noise), which may lead to a performance reduction in certain cases.
After we reduce the drives’ capacity to 32GB, the Business WinMark score grows for all configurations. As for the High End Disk WinMark score, only the Maxtor 5A300J0 installed into the STLab container and attached via USB 2.0 improved it considerably. Other configurations didn’t do the same.
The access time doesn’t play an important role here and the difference is small enough. This may be the reason for the Maxtor 5A300J0 in the STLab rack and attached via USB 2.0 to nearly catch up with the OneTouch that works across the same interface.
Now, let’s deal with NTFS. When testing the drives with their full capacity enabled, we see a predictable performance reduction (compared to FAT32) due to the lower cluster size. At the same time, there some changes in the ranks. The leader, the Maxtor 5A300J0 with the Ultra ATA interface, remained the same, though. It is much faster than the others both in High-End Disk WinMark and in Business Disk WinMark. The same Maxtor 5A300J0, but installed into the STLabs container and attached through USB 2.0, takes the second place. It has scored more High-End Disk WinMark points than the OneTouch connected via FireWire. The Maxtor B14D300 on the USB 2.0 bus is the last here.
The difference in the access time between the Maxtor 5A300J0 and the OneTouch drive is big.
When we used only 32GB of the drives’ capacity, they offered more speed. Of course, the Maxtor 5A300J0 wins the test when attached to its native Ultra ATA interface. The external drive, the Maxtor OneTouch, takes the second place when connected via FireWire. The devices on the USB bus are nearly on the same performance level.
The measured access time is nearly the same with all the drives.
The test of sequential reading and writing checks out the drive’s ability to process ordered requests with linearly increasing addresses. The request queue depth remains at 4; the size of the requested data block is changed once every minute.
Regrettably, the drive in the STLab container couldn’t pass the IOMeter tests, so we have only three results instead of five. We suspect the USB-IDE/FireWire-IDE controller employed in this case is not very perfect.
Sequential reading comes first. The diagram says there are no surprises: the Maxtor MaXLine II 5A300J0 shows its maximum speed only working via Ultra ATA. From 4KB data chunks on, its graph becomes a flat line, indicating that the drive has reached its speed limit. The results of the Maxtor OneTouch B14D300 are much lower. The transfer lines reflecting the speeds through FireWire and USB 2.0 are rising very slow; the maximum read speed is below 40MB/s with FireWire and a little higher than 32MB/s with USB 2.0.
The Maxtor 5A300J0 takes its deserved first place in this sequential writing test: it draws a graph that is similar-shaped to the one we saw in the previous test. The speed is also nearly the same. The Maxtor B14D300 is now slower than in the Sequential Read pattern, irrespective of the interface. The transfer lines are going up very slow in the diagram, but FireWire seems preferable to USB 2.0 again. This time, however, the difference between the two interfaces is small.
Now we’ve got a couple of server patterns for our drives and interfaces.
The overall performance rating is calculated as the arithmetic mean under all five workloads. As we might have expected, the Maxtor 5A300J0 with the Ultra ATA interface wins the test. The OneTouch drive is the second with the FireWire interface and third with USB 2.0. As you see, the gaps between the places are wide enough.
We calculate performance ratings for the Web Server patterns in the same exactly way as in the File Server one. Again, the Maxtor 5A300J0 is the winner; and again, FireWire is preferable to USB 2.0 with respect to the external Maxtor B14D300 drive.
The drives were working under the workload typical for a workstation, with the maximum request queue depth = 32. Again, we tested the devices in two modes: using their full capacity, and using 32GB of their address space.
The diagram contains overall performance ratings: the Maxtor 5A300J0 is the best again, faster than the OneTouch on any of its interfaces. As for the difference between FireWire and USB connections, it is negligible here.
All three configurations speed up when we use only 32GB of their storage space and there’s a wider gap between the winner, the Maxtor 5A300J0, and the OneTouch drive on any interface. The external drive profits more from a FireWire connection, rather than a USB one.
We tested the drives in the new version of our File Copy Test using our traditional five sets of files; we also created another partition of the hard disk drive, besides the main 32GB partition, so that we could measure the speed of copying between two partitions. We ran the test for two file systems, FAT32 and NTFS.
FAT32 comes first. It is beyond doubt that the Maxtor 5A300J0 works better through its native interface, Ultra ATA. The same drive takes the second place in the STLab rack and connected via FireWire. The remaining configurations are not so easily classified. Overall, it seems like the Maxtor OneTouch with the FireWire interface wins more tests over the Maxtor 5A300J0 in the STLab rack and with USB 2.0. Well, the last configuration, the Maxtor B14D300 with USB 2.0, is not at all hopeless and sometimes even takes the third place.
We see no great changes in the ranks after we switch to NTFS. Again, the Maxtor 5A300J0 with Ultra ATA is far ahead of the others. The same drive in the STLab rack and with FireWire is on the second position. The other configurations are very similar, as the external Maxtor OneTouch with FireWire loses to both configurations with the USB 2.0 interface in several tests.
The last test in this review is PCMark04 from the well-known Futuremark. We don’t always use this benchmark, which is not sharpened specifically at measuring performance of hard disk drives. Sometimes it produces questionable results, but now we decided to use it anyway. Moreover, all configurations went through the PCMark04 tests to the end.
The first test measures the hard disk drive performance during an OS boot-up. The Maxtor 5A300J0 with its native Ultra ATA interface wins here, followed by its humble self, but in the STLab rack and with FireWire. The external OneTouch drive attached via FireWire is the third. The two configurations with the USB 2.0 interface are the slowest in this test.
Once again, the Maxtor 5A300J0 with the Ultra ATA interface is an unrivalled leader at loading applications. All other configurations perform very much like each other, but you may notice that the same Maxtor 5A300J0 in the STLab rack takes the second and third places, with FireWire and USB 2.0 interfaces, respectively, while the OneTouch drive is a bit slower. Once again, we see the FireWire interface being faster than USB 2.0.
The drives were copying files of about 400MB size in this test and the Maxtor 5A300J0 attached to the mainboard’s IDE channel is on top. This same drive, but in the STLab rack, becomes the second with FireWire and the third with USB 2.0. The external Maxtor OneTouch drive occupies the fourth and fifth positions when attached via USB and FireWire, respectively.
The fourth test, General Hard Disk Drive Usage, reflects the performance of the disk subsystem when several applications are running simultaneously. Once again, the Ultra ATA-connected Maxtor 5A300J0 is far ahead of the others. Once again, it takes the second and third places when installed into the STLab rack and attached via FireWire and USB 2.0, respectively. The OneTouch drive, connected through FireWire and USB 2.0, takes the fourth and fifth positions.
The results of PCMark04 suggest that the Maxtor 5A300J0 in the STLab rack is faster than its competitor, the external OneTouch drive, probably because of the lower access time of the 5A300J0 model.
The testing session is over and its main outcome is the statement that the fastest and cheapest data reservoir of 300GB storage capacity is the Maxtor MaXLine II 5A300J0 hard disk drive. Well, I can’t say this is a surprise and you may also doubt the convenience of use of this drive in everyday work. Among the rest of the tested devices, the Maxtor MaXLine II 5A300J0 installed into the STLab rack is preferable as concerns performance, especially when it is connected across a FireWire interface. As we noted several times, it is faster than the Maxtor OneTouch drive, probably due to the lower access time of the ATA device. On the other hand, we can make this conclusion by only those tests that this configuration really passed through to the end. It is quite probable that you’ll meet similar problems with the drive installed into the STLab rack in ordinary applications, not only benchmarks. This instability is probably due to the controller of the rack, which is also very costly. Adding the price of the hard disk drive itself, I don’t think this configuration has any economical benefits over the external OneTouch device. Of course, there are numerous other containers in the market, and cheaper ones, but their stability needs testing.
The Maxtor OneTouch drive passed all the tests, without giving us a single cause to question its reliability. Its cost is even lower than that of the kit of a MaXLine II 5A300J0 drive and a STLab rack. I think these two facts – stability and price – make it a preferable buy, even despite its slightly lower performance compared to the above-mentioned kit.
As for the choice of interface in external drives, FireWire (IEEE1394) once again proved its superiority over USB 2.0 in a number of tests.