by Alexey Volkov , Nikita Nikolaichev
12/02/2004 | 11:58 AM
If you have been reading our materials carefully, you should remember that we have already offered you three parts of Mega Roundups for the drives with 80GB, 120GB and 160GB storage capacities (for details please see the following articles: Mega Roundup 3: Twenty One Hard Disk Drive with 160GB Storage Capacity, Mega Roundup 2: Twenty Two Hard Disk Drives with 120GB Storage Capacity, Mega Roundup: 20 Hard Disk Drives with 80GB Storage Capacity). Right now it looks like only extra-large drives of 200GB and bigger sizes have been left unattended. There are relatively few such discs available, so we decided not to split them further into categories (200GB, 250GB, 300GB) but to deal with them with one stroke, in one review.
We’ve collected sixteen devices in total: nine 200GB ones, six 250GB ones and one 300GB drive. The list of the manufacturers has become shorter, too. Samsung isn’t yet producing hard disk drives of such capacities, so we’ve got only four manufacturing companies you know so well: Hitachi, Maxtor, Seagate and Western Digital.
Foreseeing some remarks on your part, we want to confess from the start that this article doesn’t include the newest devices, which have become available in shops when we were writing these lines, but we promise to take a closer look at the newcomers in separate articles. Another reason to exclude the new HDDs from this roundup is their support of Native Command Queuing, and this technology is an exiting story in its own right.
Deskstar 7K250 is the only product series from Hitachi to include 200GB and bigger models. To be exact, it includes two Parallel ATA drives of 200 and 250GB capacities and one 250GB Serial ATA drive – the PATA products even share the same model name and only differ in the capacity. Hitachi must have intended the 200GB model for OEMs, not for retail, and didn’t think out a special name for it, but we spotted this HDD in retail shops.
The characteristics are rather boring – the drives all belong to the same family, are based on 80GB platters, and have 8 megabytes of cache memory on board.
Since there are no official data about the number of platters and heads for the HDS722525VLAT80 drive with 200GB storage capacity, we put the supposed data into the spec table above. All the details and explanation of our reasons for this supposition you can find in the article called Hitachi Deskstar 7K250: Vancouver 3 HDD Review.
Maxtor offers a more extensive selection of “large” hard disk drives:
The DiamondMax Plus 9 series is represented with two 200GB discs that have different interfaces. We’ve already encountered products of that line, so we are unlikely to learn anything new about them today. As for the drives from the MaXLine Plus II series, it is the first time they participate in our tests. Well, this whole series consists of only two products, but what products they are!
Launching the MaXLine II series, Maxtor defined a new class of hard disk drives which combine the good price/performance ratio of PATA/SATA devices with the highest reliability of SCSI ones. The special new design and the use of reliable components helped to boost the mean-time-to-failure parameter up to 1 million hours (against the standard 500.000 hours of ordinary desktop drives).
Thus, with all their excellent qualities, the MaXLine Plus II drives really claim to be the ideal storage media for corporative database systems (with loose requirements to the response time, of course), for the disk subsystem of powerful workstations and so on.
We have also got a real monster into our review – the 5A300J0 drive of 300GB capacity. It is the top-end model of the MaXLine II series that consists of two products (250 and 300GB capacities). Like their seniors from the MaXLine Plus II line, the MaXLine II models are intended for the so-called midline and near-line applications.
But what’s the difference between the two lines (with and without “Plus”)? Their spindle rotation speeds differ! The 5A300J0 drive is the only device in this review to have a spindle rotation speed of 5400rpm. The dissimilarity doesn’t end at that, though. MaXLine II drives are equipped with a 2MB cache buffer (against the 8MB buffer of MaXLine Plus II series HDDs) and have one platter more.
With only two products, Seagate has a smaller presence in this review compared to the previous one (see the article called Mega Roundup 3: Twenty One Hard Disk Drive with 160GB Storage Capacity for details). It’s not that we couldn’t find more Seagate HDDs of large capacities – Seagate just doesn’t produce such discs! (Yes, we are aware of the announced Barracuda 7200.8 series, but where are the actual products?)
The use of a two-platter design in the Barracuda 7200.7 series (a design that permits to install two platters at most) helped Seagate to save money on making drives of popular capacities. But at some moment it turned out that Seagate just had nothing to answer to the competitors with which were already selling 160GB and larger devices. We can imagine the indignation of Seagate’s OEM partners :). Finding itself on this Procrustean bed, Seagate was suffering Tantalus’ tortures, but they found a solution! The Seagate engineers just followed the recipe from ancient myths and “stretched” the platter to 100 gigabytes! This was the only way to reach the desired capacity of 200GB, keeping the old design at the same time.
Let’s see how it was done. The table below shows you the zone map (the distribution of the number of sectors per track into zones) of the 160GB and 200GB models from Seagate’s Barracuda 7200.7 series:
It’s perfectly clear that the extension of the platter from 80GB to 100GB was achieved not only through an increase of the number of tracks, but also through a consolidation of the tracks themselves: compare the amount of sectors per track in the highest-density zero zone!
If we were to consider the Sectors-per-Track values alone, we’d note that the platter in the 200GB drives has got four zones as an addition – the SPT value in the 4-th zone of the ST3200822A equals the SPT value in the zero zone of the ST3160023A drive. But it would be wrong to consider the SPT values only :).
Take note of the amount of cylinders in the zero and first zones of the ST3200822A – about 20 thousand cylinders of the disk have the maximum 1232 sectors-per-track density! Well, such a high data density should help Seagate’s drives in tests that demand a high linear read speed. As for the movements of the read/write heads, we can only say that Seagate continues its policy of dividing drives for “home” and “workstation” uses, which is common for the entire Barracuda 7200.7 series (see our article called Seagate Barracuda 7200.7 Hard Disk Drive Family for details). In other words, the heads of the ST3200822A are slowed down for noiseless work, while the ST3200822AS, on the contrary, has fast heads for the maximum performance. You cannot control the Acoustic Management system of these drives, so you have to put up with it.
The question of support of Native Command Queuing technology by the ST3200822AS drive was long arguable, opinions dividing. Our attempts to “find” NCQ in our ST3200822AS sample bought in a retail shop were unsuccessful.
In fact, Seagate can anytime equip the ST3200822AS with an electronics board that supports Serial ATA II: Extensions to Serial ATA 1.0 Specification, as they did with the junior SATA models of the Barracuda 7200.7 series. But we think that Seagate will not do that. Why? Read the next table:
We think that Seagate will just force the ST3200822AS out of market, replacing it with the drive from the Barracuda 7200.8 line, because the latter has better characteristics and uses the standard components of the 7200.8 series.
As the result, Seagate’s product line-up will become complete – 7200.7 series drives will fill the low-end sector (the SATA models will be updated with new electronics), and the 7200.8 series will compete in the top-end sector.
So, for the better or for the worse, the ST3200822A/AS models have done their job and can leave now…
Western Digital, as usual, offers a variety of hard disk drive models.
There are four discs of 200GB capacity present. We met the “BB” and “JB” models before, but “PB” and “JD” are something new. Let’s deal with them one by one.
WDxxxxPB are new hard disk drive models with the PATA interface, an 8MB buffer and fluid bearings. These drives add low-noise products to the high-performance Caviar series (with the “JB” suffix). It’s quite logical, considering the numerous complaints of the happy owners of “JB” drives at their noise at work. To reduce the noise, the “PB” drives use fluid bearings (by the way, it seems like Western Digital is steadily transferring all its drives to such bearings) and bigger delays between movements of the heads (the drive comes with AAM = 128).
WDxxxxJD are high-performance disks with the Serial ATA interface. This model can’t boast any exceptional characteristics, as it is in fact a replica of the PATA “JB” drive. To add the interface, a PATA-SATA converter from Marvell is employed. There’s nothing wrong about that – all manufacturers, except Seagate, made their first generation of SATA drives with the help of a PATA-SATA bridge.
So, we have mustered all the participants of our today’s tests. As we mentioned above, Samsung didn’t tinker with 100GB platters, preferring to jump over right to 120GB ones. That’s why you won’t see Samsung’s drives in this review.
We have to use two controllers as we have hard disk drives that connect across two different interfaces. We took two controllers from the same manufacturer:
The testbed was configured as follows:
The drives had the following firmware versions:
We used the following benchmarking software:
Note that unlike in the previous reviews, we use a new version of FC-Test now (see our article called X-bit's FC-Test 1.0 or "System Rebooted"). Thanks to the system reboots after each test action, this version eliminates the effect of caching of files in the system RAM and thus yields more correct results.
We wrote the Maxtor drives through before the tests to avoid their forced write verification. We turned off the noiseless seek mode for the WD2000PB drive.
For WinBench tests we formatted the drives in FAT32 and NTFS as one partition with the default cluster size (FAT32 formatting was performed with Paragon Partition Manager). We ran the tests seven times each, chalking up the best result. The HDDs didn’t cool down between the tests. For FC Test we created two 32GB partitions on the drives. For IOMeter tests we used Sequential Read, Sequential Write, Database, Workstation, File Server and Web Server patterns. You can refer to our previous reviews for details about the patterns.
The Database pattern traditionally opens up the show. This pattern helps us check the drives’ performance with a stream of requests to read and write random-address 8KB data blocks. Changing the reads/writes ratio we can evaluate the quality of sorting the read and write requests.
Since the table is too large, we have to divide the results of this pattern into groups by the manufacturer.
For better readability, we draw graphs for different loads (the number of simultaneous read/write requests).
The two PATA devices behave in the same way, but the SATA model is quick to lose its ground. It definitely doesn’t like write requests. Curiously enough, the PATA and SATA drives from Hitachi are behaving absolutely differently in this test, quite contrary to what we saw in our tests of 160GB HDDs where drives with different interfaces, but with the same amount of cache memory, had the same speed in all the modes. Here, the SATA drive evidently has a less aggressive lazy write algorithm, and this is in fact easily explained: top-end SATA-interfaced products can be positioned as “hard disk drives for entry-level servers”, and aggressive lazy writing isn’t welcome for such an application.
The general picture has remained the same at request queue depth = 16. The SATA drive is much slower in all the modes where write requests are present.
Under the maximum load the SATA disk is just a little slower than the two ATA drives, which are still performing much alike to each other. The highest performance of the PATA models in the Random Write mode needs a separate mention: it is the first time drives with a spindle rotation speed of 7200rpm reach a speed of 200 processed requests per second.
The most curious fact, to our opinion, is that the SATA and PATA models from Hitachi behave differently, in spite of their having the same firmware version.
The Maxtor team comes next:
And the diagrams:
We could have predicted beforehand that the 5A300J0 would be the slowest drive in our tests. The low spindle rotation speed and the small amount of onboard cache memory don’t allow it to compete successfully with the rest of the drives from Maxtor. As for the other HDDs, they have almost the same speed.
The 300GB drive is still lagging behind under the medium load. The 7Y250M0 model falls out of the common group in modes where read requests prevail, and the 6Y200P0 does the same at high percentages of writes.
There are almost no changes under the maximum load – the results of the drives have become even closer. Let’s note once again that there’s no distinct correlation between the firmware version and the performance of a drive and move on to Seagate’s products.
We’ve got just two drives from Seagate, so analysis is going to be an easy job. :)
Under the linear load (1 request in the queue), the ST3200822AS is head above its ATA-interfaced mate, although the two graphs drew nearer at 100-percent writing. Well, as you remember, the advantage of the SATA drive over the PATA model had been "planned" originally (see this article for details), so let’s not make a face as if we were in any way surprised.
Let’s see if the same goes for higher loads.
The drives both accelerated under a load of 16 requests when performing reads. The ST3200822A is slightly slower than its SATA mate under mixed loads, but is getting closer to it at high percentages of writes.
The discs perform more similarly under the maximum load. At 100% write requests, the drives deliver the same performance. At 100% reads, the ST3200822A is just slightly slower than the other drive.
That’s a queer diagram as the drives from Western Digital are reading in the same way under the small load, but diverge in a fan-like fashion as soon as there appear write requests in the queue. The WD2000PB was the best of all, and the WD2000JB was the worst. What’s characteristic, all drives with the same interface have identical firmware.
The overall situation remained the same under the 16-requests load, but the WD2500JD fell somewhat behind at reading. The two drives with the Serial ATA interface are behaving in the same way, but the WD2000JD, which has a smaller capacity, slowed down on 100% writes.
The graphs have become more concentrated under the maximum load – it’s hard to determine the leader. At high writes percentages, the WD2000PB and the WD2500JB are better than the rest of the team. The ex-outsider, the WD2000JB, is getting closer to the common group.
We use IOMeter to send a stream of read/write requests to the drive with request queue = 4. Once a minute the size of the data block changes, so we can see the dependence of sequential read and write operations on the size of the data block.
Click to enlarge
The Maxtor 5A300J0 is the best at reading small blocks as it reaches its maximum speed on 4KB data chunks already. Seagate shows the power of its 100GB platters on large data blocks as the drives from that company overcame the 60MB/s mark. On the other hand, these drives are the slowest on small data blocks (up to 8KB).
The slowest drive on large blocks is the WD2000JB – we didn’t retest a newer sample of the same model, but took an older sample.
It wouldn’t be convenient to put all the graphs into one diagram, so we split the participants into groups according to the manufacturer. You can view the diagrams by the following links:
Now let's turn to sequential writing:
Click to enlarge
The Maxtor 6Y200P0 is the fastest when writing small data blocks, but the other Maxtors show good results, too. The Seagate ST3200822A is again the best with large data blocks, while the other Seagate (with the SATA interface) is slower (due to a different cache segmentation?), but keeps its second place.
You can view the diagrams by the following links:
These patterns simulate the load on the disk subsystem of a typical server.
We compare the drives in this test by their performance ratings, which are calculated as the average of the drive’s speeds under four loads (we don’t count a load of 256 requests in).
We’ve got two leaders here – the 200GB Hitachi HDS722525VLAT80 and the Seagate ST3200822AS. The gaps between the other drives are negligible.
The Maxtor 5A300J0 is very slow, but don’t forget that it is the only HDD in this test to have 5400rpm spindle rotation speed. By the way, none of the other Maxtor drives can boast a high performance in this test, either. Let’s see how they handle the Web Server pattern?
The three drives from Hitachi populated the top of the diagram. The 200GB Hitachi HDS722525VLAT80 is especially good in this test. The SATA-interfaced drive from Seagate is trying to challenge the Hitachi team, while the PATA drive from Seagate found itself at the bottom of the table, before the Maxtor 5A300J0.
This pattern has a big share of write requests, and this condition should bring some changes into the rankings.
Well, the test is different, but the leaders are the same – the 200GB Hitachi HDS722525VLAT80 and the Seagate ST3200822AS. The WD200PB and WD200BB drives from Western Digital took the third and fourth places, ousting the Hitachi HDS722525VLAT80 to the fifth. The Maxtor 5A300j) is the slowest device, again.
Now we reduce the storage space of the drives to 32GB and repeat the test.
The 200GB Hitachi HDS722525VLAT80 keeps its leading position, but the “full-size” HDS722525VLAT80 is very close behind. The WD2500JB is on the podium, too.
This is the end of synthetic IOMeter tests – we can proceed to WinBench now.
We use WinBench to check the hard disk drives in the “desktop PC” mode. We format the disk in the NTFS file system by means of the system tools (the default cluster size is 4KB) and in FAT32 using Paragon Partition Manager (the cluster size is 32KB). We also perform our tests on 32GB storage space in NTFS and FAT32 file systems (partitioning the drives be means of the standard Windows 2000 Disk Manager).
We will first discuss the physical parameters of the HDDs, which don’t depend on the file system.
The 200GB HDS722525VLAT80 model from Hitachi has the best average access time – such a big advantage over it own mate makes us reconsider the idea of a five-platter design in the 200GB drive. But who knows…
The Seagates perform quite as expected – the ST3200822AS model takes the third place, and the PATA model resides on the last line of the diagram.
The results of this test are indicative of the correlation between the areal density and the linear read speed.
That’s right – the 100GB platters win this test for the two drives from Seagate. Other HDDs have similar results among themselves with the exception of the 200GB drives from Maxtor which seem to be made on “shortened” platters, judging by the high transfer rate at the end of the disk.
You can view the data-transfer graphs of the drives by the following links:
We will watch the drives work in the FAT32 file system first.
We’ve got a lot of data, so we split the table of results into several parts:
Let’s discuss the results in two integral tests: Business Disk Winmark and High-End Disk Winmark.
The Maxtors took the top four positions in the High-End Disk Winmark subtest – a pair of SATA drives is followed by two PATA devices. The two full-size drives from Hitachi take the fifth and sixth places. As for the Business subtest, the results don’t vary much here, save for the two outsiders (WD 2000BB and Maxtor 5A300J0). What’s notable, these two outsiders are equipped with a 2MB cache buffer.
Now, let’s run these tests again in NTFS.
You can view the full data by the following links:
It’s almost the same as in FAT32, only the speeds are overall smaller and the two pairs of Maxtor drives have exchanged their places (once more we see a clear correlation between the test result and the firmware version with Maxtor’s HDDs).
Again, the Maxtor team wins the High-End subtest and the Hitachi team is superior in Business Disk Winmark.
We decrease the storage space to 32GB and run the tests once again.
And again there are only minor changes: the drives from Maxtor are in the lead, although in a different order. The Seagate ST3200822AS is faster than the 200GB Hitachi HDS722525VLAT80.
The same test, but in NTFS:
Like in the previous case, there are no surprises here. The Maxtors took the top of the diagram for themselves, and the 200GB HDS722525VLAT80 regained one position.
Next goes our most interesting benchmark – File Copy Test.
We use FC-Test according to our standard methodology. We create two 32GB partitions on the drive and format them in NTFS and then in FAT32. Next we create a set of files on the first volume, read it from the drive, copy it into a folder on the first partition (i.e. we copy files within one partition), and then copy it into another partition. FC-Test 1.0 build 11 has one important improvement over the older version – now the utility reboots the testbed after each test action, so we prevent the results from being influenced by the file caching performed by the operating system. We use five file patterns:
Let’s start with the NTFS file system. Because of the so many drives included, we will be discussing the results by each action for each file pattern. The first action is the creation of a file pattern on the disc.
The following diagram shows the speed of the creation of the Install pattern:
The Maxtor drives occupied the top four positions, like in the WinBench tests. Behind the leading group, there’s tough competition going on: the Hitachi HDS722525VLAT80 and the Seagate ST3200822AS have almost identical results. The Maxtor 5A300J0 and the WD2000JB grappled with each other – youth won.
Let’s see how these results change in ISO and MP3 patterns that consist of bigger files.
The drives from Maxtor are beyond competition as concerns creating big files. The drives from Hitachi sank to the lower part of the diagram. The two Seagates are firmly seated in their fifth and sixth places, while the slow Maxtor 5A300J0 is successfully challenging the drives from Western Digital.
The remaining patterns – Programs and Windows – consist of many small files.
The Maxtors maintain their advantage at creating small files, but the Hitachi team is close behind them. The drives from Seagate are rather bad at such kind of work: the ST3200822A plummets to the bottom of the diagram in the Programs pattern and the ST3200822AS does the same in the Windows file-set.
That’s enough for creation, let’s do some reading:
When reading the Install pattern, the two Seagate HDDs wrest the leadership from Maxtor and Hitachi due to their advantage in areal density or, maybe, due to their firmware algorithms.
Reading long files is the ideal workload for Seagate’s HDDs as they can profit by their advantage in areal density – and they do! The other drives are reading the ISO pattern at about the same speed, except two outsiders – Maxtor 5A300J0 and WD2000JB.
The Maxtor 6Y200M0 manages to outperform the Seagate HDDs in the MP3 pattern, and three more Maxtors have forced the Hitachi drives down in the diagram. The drives from Western Digital are hopelessly slow.
Reading files of the Programs pattern shows that Seagate’s drives are handling smaller files well, too. The drives from Maxtor are suddenly slowing down here, but manage to remain faster than the products from Western Digital.
The drives form two groups when reading files of the Windows pattern: the worst group includes all the drives from Western Digital and the slowest Maxtor (5A300J0), and the best group is very dense. We don’t think that we should single out any particular model here.
Next goes copying into a subfolder on the same partition.
As you see, the drives from Western Digital deliver the worst performance, again.
When copying files of the Install pattern into a subfolder, the two disks from Maxtor keep their advantage, but now they are closely followed by the two ATA-interfaced drives from Hitachi. The drives from Seagate are not as good at copying as they are at reading, so their results are average here.
The Maxtor drives feel even more confident when copying large files. The drives from Hitachi are far behind the Maxtors in the ISO pattern, but approach them in the MP3 file-set.
The copying of small files allows the Hitachi team to take the podium, while the Maxtors are outperformed by the Seagate drives.
Now, the last action of FC-Test is copying files to another partition.
Copying Install files to another partition has put the drives in the alphabetic order almost perfectly, save for Maxtor being later than Seagate in the diagram.
Once again the Maxtor drives are the best at copying large files (of course, save for the 5A300J0 which is at the bottom of the diagram). Hitachi’s HDDs are grouped after the fast Maxtors; Seagate and Western Digital are competing in the next tier.
When copying the MP3 pattern, the Maxtor 7Y250M0 and 7Y250P0 are on top, followed by the Hitachi HDS722525VLAT80. Seagate’s HDDs have got closer to the leader.
Copying small files of Programs and Windows patterns brought victory to the Hitachi team again, but the Seagates did well, too. The drives from Maxtor are not very skilful with small files, as we have already seen in earlier tests.
That’s the end of tests in NTFS. The results for the FAT32 file system are similar to those we saw under NTFS and we don’t think it necessary to discuss them at length, just a couple of facts: first, the speeds are overall higher in FAT32 than in NTFS, and, second, the drives from Maxtor work somewhat better in NTFS than in FAT32, so the Hitachi and Seagate drives are more often in the lead when copying large files in FAT32.
Here are the results and the diagrams for FC Test in FAT32:
Once again we cannot name an absolute champion of our tests. Different drives took the podium in each of the benchmarks and we think it incorrect to normalize the results to the winner of each test and sum up these coefficients to determine the overall winner by the total score. We’ve got very different tests, and there are few users who are using their drive as the disk subsystem of a server and as the disk for processing streaming video at the same time…
So, we will only name the winner in each subtest for you to choose what’s required for your particular needs.
The Hitachi HDS722525VLAT80 drives of 200GB and 250GB capacities and the Seagate ST3200822AS are the best in the IOMeter tests that emulate the disk subsystem of servers. Thus, these drives are going to be a good choice if you’re building a simple server.
The Hitachi team, the SATA model from Seagate and two drives from Western Digital, WD2000PB and WD2000BB, were the best in the patterns that emulate the disk subsystem of a desktop computer.
The four drives from Maxtor were unrivalled in the High-End Disk Winmark subtest of WinBench 99, and the Hitachi team won the Business Disk Winmark subtest, although with a very small advantage over the others.
It’s hard to determine a leader of FC Test: Seagate’s drives are the best at reading files, irrespective of their size; the Maxtor team is speedy at writing file sets and is good at copying large files; the Hitachi drives excellently handle small-size files.
And here are a few general observations:
So, the HDD-making technology is developing, but we’re not marking time, either. And soon you’re going to see tests of drives with Native Command Queuing on our site.