by Nikita Nikolaichev
11/06/2003 | 11:26 PM
When I started working on this roundup, I was a little bit confused. About a year and a half ago I was sincerely surprised and kept asking myself: who could ever need a 120GB hard disk drive (see our article called 3 IDE Hard Disk Drives with 120GB Storage Capacity)? And today the maximum storage capacity of hard drives with 7,200rpm spindle rotation speed has notched the impressive value of 250GB. Some time ago when Seagate marketing people were pushing forward the new Barracuda 180 they decided to talk about the storage capacity comparing it with a pile of writing paper, which could be needed to print out all the documents stored on this drive. Of course, this comparison was in favor of the drive, because the pile could be literally as tall as some well-known skyscrapers.
Today the storage capacity of much simpler and better value IDE HDDs (built of three platters instead of 12) is so high that you will simply break you neck if you try to see to the top of the paper pile equivalent to the 250GB of stored data…
In fact, do we really need such big hard disk drives? The arguments for the positive answer in case of some corporate tasks are very easy to find, and what about the home usage of these monsters?
In fact, the demand turns out to be right there.
About a few days ago I decided to rearrange my modest MP3 files collection and suddenly discovered that the average space required for a single music album is about 100MB (Lame, 320 bitrate). Having extrapolated my collection to “beyond modest” size, I realized that my current 80GB hard disk drive will not be big enough to store two operating systems with all necessary utilities, a digital photo gallery, some MP3 music and a few dozens of games.
Of course, many of the Above listed things will hardly be called really necessary ones, but it is so hard to delete anything, you know? The passion for collecting things is somewhere deep inside me, even deeper than genes :) According to some astrological magazine I used to be a relic keeper in some Borneo tribe :)
So, the human weaknesses will make large hard disk drives always popular, and the only question is how quickly the demand for high storage capacity will actually grow.
Having analyzed my own HDD purchases, I managed to derive a certain rule, which I subconsciously followed when I was upgrading my system: every new HDD I bought was three times as big as the previous one. It allowed moving the contents of the previous HDD to a new one (as I have already told you, I do not like to lose data), building up a new system on the drive and there still was quite a bit of space left on the drive to last me another year or so.
According to this rule, my next HDD should be 250GB big, which returns us to the eternal question: how do I choose the right HDD with such big storage capacity?
At present there are only three companies, which offer 250GB HDDs with 7,200rpm spindle rotation speed. They are Hitachi, Maxtor and Western Digital. By the way, the latter was exactly the one to open the era of a quarter-terabyte hard drives with 7,200rpom spindle rotation speed. As you remember, this spring Western Digital announced its WD2500JB HDD (see our article called Western Digital WD2500JB HDD: More than Drivezilla?!). Closer to the fall Maxtor MaxLine II HDDs started appearing in the mass market and not so long ago we managed to also get hold of Hitachi 7K250 solution.
Why don’t other hard disk drive manufacturers, such as Seagate and Samsung, offer HDDs with this attractive storage capacity? Especially since both these companies have 80GB platters at their disposal… As for Samsung, we know only one thing about their strategy in this market: the market will be conquered by 2005 :)
As for Seagate, we do know a little bit more about them. As is known the company has been sticking to dual-platter design for the desktop drives for the last few years. It allows reducing the production expenses for HDDs with low and medium storage capacity, which is very important for a company with a lot of orders. At the same time dual-platter design limits the maximum HDD storage capacity to 160GB, if the standard 80GB platters are used. Since Seagate did realize, how badly the market needed high capacity drives, they undertook a really unprecedented move: they added a 200GB drive to their 7200.7 HDD family. As you may have already guessed it is built of 2 platters 100GB each. These platters are an industry record today, but as you may also see it does not fit into the standard capacity scale: 20GB – 40GB – 60GB – 80GB. During the past years when the new HDD generation arrived the platters storage capacity grew 1.5-2 times higher. And the storage capacity of the new Seagate 100GB platter is only 25% higher than the industry’s standard 80GB platter. It means that 100GB platter is not the second generation platter for Seagate. Another thing proving this point is the fact that Seagate announced this 200GB HDD as a new model within the current Barracuda 7200.7 product line.
I had a very interesting conversation with a Seagate representative at the press conference on October 23, 2003. And he actually did confirm all the above described suppositions of mine. But, it is not the right moment to raise the curtain of mystery yet :)
Let’s return to our HDDs. In the today’s roundup we will compare 5 hard disk drives with 250GB storage capacity. Three of them support ATA interface and the remaining two – SerialATA interface. So, please meet our testing participants.
Hitachi 7K250 HDD boasts as sophisticated marking as IBM drives, for instance. Just look at that: HDS722525VLAT80.
You should read this according to the following rules:
S =Standard (vs A for Auto for example)
72 =7200 rpm
25 =full capacity =250GB
25 =model capacity =250GB
V =generation code
L =low profile (1-inch form factor)
AT =ATA (or SA =SATA)
8 =8 MB buffer (or 2 =2 MB)
But nevertheless, it still looks very familiar:
Hitachi 7K250 (HDS722525VLAT80)
Maxtor MaxLine Plus II (7Y250M0)
Maxtor MaxLine Plus II (7Y250P0)
The interface of the two Maxtor drives is marked with the last but one letter of the model name. This way 7Y250P0 supports ATA/133 interface and 7Y250M0 supports SerialATA interface.
WD hard disk drives have the supported interface type marked with the last letter of the model name: 2500JB supports ATA/100 interface, while 2500JD is a SerialATA drive.
Western Digital WD2500JB
Western Digital WD2500JD
These are all our testing participants for today. Now I have to say a few words about the testing methodology to be used, because it will differ a little bit from what we are usually applying.
Well, trying to meet numerous requests for a testbed upgrade we did our best and today we are proud to introduce our new test system to you:
We still used the same tests for our investigation:
However, we are going to use them in a bit different way this time. If you are following most of the hard disk drives reviews and benchmarks, you may have noticed that many web-sites and offline magazines offer a more or less standard similar pattern set for Intel IOMeter and standard approach to WinBench. Therefore, I thought it would make sense to make a few changes to our hard disk drive testing methodology.
We will be testing every hard disk drive in WinBench99 for each file system: in a 32GB partition (when the logical drive is created in the beginning of the disk) and in a partition as big as the whole drive. All WinBench99 tests were run 7 times each and then the best result was taken for further analysis.
The second innovation in our testing methodology has to do with WorkStation pattern. In fact, the pattern itself remained the same, but the testing approach has become different. To evaluate how big the “locality” influence on the HDD speed is, we ran the pattern twice: on the entire hard drive and on the first 32GB of the drive.
The third innovation touched upon the FC-Test. Now I am not just cutting the drive into two parts of equal storage capacity but I am creating two partitions in the beginning of the drive with 32Gb storage space in each.
The synthetic patterns for Intel IOMeter and the server patterns remained unchanged. The HDDs didn’t cool down between the tests.
The testing participants had the following firmware versions:
As usual we are going to start with the DataBase pattern. If you are curious about the exact numbers obtained during the tests, please have a look at the table below, which carries the results for Total I/O under five types of workload (we changed the requests queue depth) in 11 modes with different writes share (from 0 to 100% with 10% shift).
Now let’s have a look at the HDD performance during requests processing under three key workloads:
Under linear workload (one outgoing request) our hard disk drives have already proven individual enough, so that we can start drawing some conclusions.
Of course, the Hitachi drive feels completely at home in the marginal modes. In both: RandomRead and especially RandomWrite, it proved much faster than the competitors. At the same time, it doesn’t seem tom like those cases when the reads and writes are more or less balanced.
Maxtor drives are faster than the competitors from Western Digital in RandomRead mode, but yield to them in all other cases.
As for the performance differences between the ATA and SerialATA drives from one and the same manufacturer, the situation here is very interesting. The SerialATA drive from Western Digital appeared faster than its ATA counterpart, while by Maxtor the situation turned out just the opposite: the ATA HDD outperformed the SATA one.
Now let’s find out what happened with 16 outgoing requests.
Under this workload the performance difference between the drives is more evident, however, the graphs overall shape remained unchanged.
When the workload increases to the maximum of 256 outgoing requests, we can see that the two Maxtor graphs would merge into one. Hm… Despite different firmware versions of the ATA and SATA hard drives, they behave absolutely identical in DataBase pattern. Hitachi drive managed to break all the records in RandomRead and RandomWrite modes, though in intermediate modes it appeared as fast as any of the Maxtors.
Western Digital hard drives proved to have the best balanced algorithms: their graphs are very straight and they indicate a clear performance growth with the increase in the writes share.
In fact, there is one more thing that can be seen from the DataBase results. Since this pattern works with 8KB data blocks, which size hardly matters for the today’s high density HDDs, we can regard the requests processing speed for queue=1 as random access time value. The performance difference during 8KB data blocks reading will make 0.1ms compared to 512Byte blocks reading speed, which can be neglected. This way, if we take Total I/O results for RandomRead and RandomWrite modes in case queue=1, we will be able to calculate the access time during reading and writing. And if we divide the former by the latter we will get the lazy write algorithms efficiency coefficient.
Please pay attention to the Hitachi’s results. It is the first time that I remember of when the lazy write efficiency coefficient exceeded 2! This is a really great job!
Intel IOMeter software sends to the HDD a set of reads and writes with the queue=4. Once per minute the data block size changes that is why in the end we can actually track the dependence of the linear read and write speed on the data block size.
To make the results table more illustrative I highlighted the best results with blue and the worst ones with read color for each data block. However, you will see the HDDs performance difference even more clearly on the graphs:
Of course, Maxtor hard drives are the fastest when it comes to smaller data blocks. Note that the SATA WD hard disk drive is as fast when working with small data blocks as Maxtor solutions. However, the parallel ATA Maxtor HDD is considerably better than both of them.
As it comes to larger data blocks Hitachi drive dashes forward. Starting with 16KB data blocks the HDD runs at its maximum speed and then doesn’t matter on the data block size any more. And as for WD 2500JD HDD, this seems not to work for it: for some reason it didn’t like 256KB and 512KB data blocks. I also have to say that WD hard disk drives are slower on large data blocks than the competitors’ solutions.
Now let’s check how well they cope with writing:
During writing none of the HDD suffers any performance drops, except WD’s “brand name” confusion with 1KB blocks (the confusion is caused by the fact that the minimum WD cache segmentation is only 1 sector).
The winners here appeared two Maxtor solutions. They are considerably faster than the others as it comes to small data blocks and a little bit ahead of the rivals on bigger data blocks.
Well, it’s high time we passed over to server patterns now.
The table with highlighted good and bas results comes first:
Of course, the indisputable leader here is the Hitachi solution. In WebServer pattern is it simply beyond any competition (remember how fast it was in RandomRead mode). However, in FileServer pattern both WD drives compete with the one from Hitachi.
And this seems to be a really cut-throat competition, though only under high workloads.
Hitachi drive also appears very high up in the rating list. Note that ATA and SATA HDDs from Western Digital are equally fast here, while the two drives from Maxtor are no longer that unanimous: the SATA HDD turned out somewhat slower.
When we look at the WebServer performance graph, the advantage of Hitachi drive over competitors appears really striking.
And of course, this can’t help telling on the rating.
I am pretty excited to begin the discussion of WorkStation pattern results. First, we are going to see for the first time how the work within a “limited disk space” is going to affect the current ratings of the drives. You should remember that I decided to test the drives within the first 32GB of their disk space this time. Because we are primarily interested in the HDD performance in the beginning. This is where we usually have the OS, the paging file and a number of other frequently used files.
Second, the HDDs performed in such an interesting way in DataBase pattern that I turned out unable to make any forecasts at all (this is the first time that happens to me, honestly).
So, what is typical of this pattern? Of course, it is the high share of writes, which should, which is supposed to help WD drives do better, but… we shouldn’t disregard the brilliant performance of Hitachi solution. Anyway, I am not going to keep these exciting results top secret any longer. Here you are:
The results of our testing participants in the regular WorkStation pattern are simply amazing! The leader is the Hitachi drive! Actually, I expected WD solutions to be indisputable leaders with such a high share of write requests, which we have just seen in DataBase pattern.
However, WD drives manage to compete with Hitachi only under really heavy workload (for this pattern, of course).
Maxtor drives with different interfaces again appeared very diverse. Moreover, the SATA hard disk drive from Maxtor turned out the slowest of all, while the Parallel ATA drive proved to be pretty competitive.
Although the graph above indicates that Hitachi is not that far ahead of the rivals, really.
The rating shows that it again outpaced WD drive by just a few percents. By the way, the Parallel ATA drive from Maxtor appeared the third, having got just in between the WD pair.
And as for the tests in the first 32GB of storage space, they are more than exciting.
Look at the gap between the Hitachi drive and the rivals! This is an excellent job! But what determined this victory: great lazy write algorithms or smart command queuing?
Since the heads moving distance has become considerably smaller, the Hitachi drive appeared not the only one to benefit from that: both WD solutions also defeated the Maxtor HDDs completely.
This is what the situation looks like during the tests on the first 32GB of storage space. Hitachi hard disk drive is far ahead of the competitors here, while both Western Digital solutions and the Parallel ATA Maxtor HDD run neck and neck.
We also use WinBench test set to check the HDD performance in applications. As you remember, WinBench tests are performed on a freshly formatted drive, which leads to a certain dependence of the benchmark results on the file system parameters. According to the traditional testing methodology I tested the drives formatted as a single partition. Therefore, if the drive features big storage capacity, the amount of auxiliary info to be stored on it also increased. So, when I was comparing the results for HDDs from the same product line but with different storage capacities, I noticed one very interesting tendency. As the HDD storage capacity increased, its performance in WinBench test started growing at first and then went down. The performance increase is easily explained by the fact that the more platter surfaces are involved, the smaller is the heads moving distance, when you need to read or write the file. The performance drop, however, turned out much more difficult to explain. I believe that it has mostly to do with the growing “expenses” of the OS when working with high capacity drives. But let’s stop guessing: we will have clear evidence in a few minutes.
We will start with FAT32:
Firstly, I would like to draw your attention to the access time of Hitachi HDD, which is equal to 11.8ms! Wow, I haven’t seen numbers like that for ages, I suppose. At least since the times of Quantum LM HDD.
This is an excellent result, I should say. And the remarkable thing about it is the fact that this positioning speed can be achieved on 83GB platters! It definitely means that there is a way to combine high storage capacity and fast mechanics in a single drive.
To make the results a bit easier for you to read, I summed up the performance values in two integral tests (namely, Business Disk WinMark and High-End Disk WinMark) for the entire drive and for the first 32GB of drive space. The picture appeared really interesting:
As we see, all HDDs perform faster within smaller partitions rather than in case the entire HDD space is involved. This way, the theory I have shared with you got a point to its score.
If we forget about all theories for a while and compare the HDD performance as is, we will see Hitachi HDD again ahead of all. Although this time WD solutions will be really close behind it. Maxtor HDDs were the last ones this time.
I wonder if the partition size tells on the results in the High-End test:
Of course it does! And look how big this influence actually is! But only not for WD drives :)
It is pretty interesting that Maxtor drives got really fast here: their performance is much better than that of the competitors. The third prize was taken by Hitachi, while WD drives keep going side by side, as usual.
I wonder if the same situation will repeat in NTFS file system:
Hm, on the one hand we see the same tendency here, but on the other, it is not quite the same…
It turned out that the performance of the first three hard disk drives is almost the same in a 32GB partition and in the 250GB partition. The Western Digital drives, however, do have very different performance results.
Just like in FAT32 the first prize belonged to Hitachi HDD, and the second and third one were taken by two Maxtor drives having made up for the failure in FAT32 benchmarks this way.
In High-End tests Maxtor drives are simply beyond any competition, especially the model supporting Parallel ATA interface. The third one is Hitachi, while WD HDDs are the slowest of all. In fact, it is really hard to say why they turned out so slow in NTFS…
Before we pass over to the FC-test results discussion, I would like to offer you the linear read speed graphs, which you might consider interesting to check out:
The last benchmark we are going to look at today is FC-Test checking the drives’ performance during file creation, read and copy. Please check out our article called X-bit labs Presents: FC-Test for Hard Disk Drives. And before we start the actual discussion, I would like to point out that we didn’t format the drives as two partitions of the same storage capacity, but simply created two 32GB partitions (in the beginning of the HDD) and ran all the tests within these partitions.
Well, Maxtor drives are beyond any competition during file creation. In fact, there is nothing to be surprised with: remember how fast these drives were in SequentialWrite pattern.
Hitachi HDD appeared the second fastest. It lost to WD only once, when we were creating files from the ISO set. If you remember WD drives love big files, while IBM/Hitachi drive simply hates them.
However as it comes to reading, Hitachi drive outperforms Maxtor in all file sets. And the latter in its turn is stably ahead of the WD drives.
During file copy within one partition Hitachi drive manages to win in three file sets out of five losing to the rivals only when the file size is pretty big. However, look at the advantage it demonstrates on small files! Excellent read ahead and lazy write algorithms show their best!
File copy to another partition doesn’t change the situation.
Maybe the shift to a different file system will affect the results?
Well, at least it doesn’t happen during file creation. Maxtor drives are still faster than the rivals.
And during file reading the changes are more than evident. As you remember Hitachi drive was a leader in FAT32. here Maxtor managed to dash forward.
And during file copying the situation is very similar to what we have already seen in FAT32: Maxtor drives are faster when processing large files, while Hitachi is the best when working with smaller ones.
Well, we have just successfully completed the tests of the five world’s largest hard disk drives with 7,200rpm spindle rotation speed. So, what conclusions can we draw then?
First of all I would like to say that no matter how big the HDD storage capacity is they still remain typical products of the companies which developed them. The benchmarks results revealed all family traits of these solutions, which allow us to see evident connection between Maxtor MaxLine Plus II HDDs and DiamondMax Plus 9 HDDs. The same is true for the drives from Western Digital.
I would also like to say that SerialATA interface is still no guarantee of the hard disk drive fastness. Probably the type of workload during our test session doesn’t allow us to feel the advantages of the SATA interface. Moreover, all contemporary HDDs except for Seagate drives SerialATA support is implemented via the PATA-SATA bridge, so you shouldn’t even hope for something remarkable then. The only thing that we can hope to get from the SATA support implemented in HDDs of large storage capacity is simpler integration of the drives into DataBase servers.
Of course, the indisputable star of our today’s test session appeared Hitachi hard disk drive: excellent seek time and interesting algorithms implementation helped this solution to win in many benchmarks. I would like to specifically stress its extremely high performance in server patterns.
Maxtor drive, on the contrary, didn’t show anything extraordinary in server patterns, but coped perfectly well with streaming reading and writing requests. Which we have just seen in read, write and file copy tests. This way, Maxtor solution would become a good choice for a multimedia PC, which often has to process a lot of large files. Also if you are about to assemble a video server, I suggest that you consider getting this HDD for such a system.
Western Digital hard disk drives proved pretty fast in server patterns, but their actual trump is the desktop field. So keep that in mind when shopping.