To begin with, we would like to introduce to you the world's largest hard disk drive - Seagate Barracuda 180. Here it is:

At first sight, the most peculiar thing about this HDD that catches your eye is its height: it is 1.5 times higher than the ordinary 3.5" hard disk. On the photograph below you can see IBM DTLA HDD lying side by side with Barracuda 180. Feel the difference!

Please take a closer look at the side ribs of Barracuda 180 HDD: they improve the cooling of the drive and make its case very strong and durable.

Seagate Barracuda 180 is available with three different types of interfaces (Wide, SCA and FC) and with two versions of cache-buffer (4MB and 16MB).

During transportation the HDD is protected against mechanical and electrical damages by a special plastic skin called SeaShell. And during operation G-Force Protection and Rotational Vibration Protection systems are responsible for proper functioning and security of the hardware piece. Also the hard disk drive is equipped with a thermal diode, which reports the cases of HDD overheating to the host-controller.

Seagate Barracuda 180 is provided with a special SAMS (Seagate's Advanced Multidrive System), which optimizes performance in server and RAID applications. Unfortunately, we didn't manage to see it working because we had only one Seagate Barracuda 180 HDD at our disposal.

As it follows from the name of the disk drive, the HDD rotates at 7,200rpm and boasts an incredible capacity of 180GB. The previous record equal to 73GB, which was set by Seagate Cheetah 73, Seagate Cheetah 73LP and Quantum 10K II, has now more than doubled. It turned out possible due to a combination of 12 platters 15GB each. Note that the overall height of the HDD got only 1.5 times bigger. We should give credit to Seagate engineers who managed to stuff this relatively small case with 12 platters and 24 read/write heads! Moreover, the platters used in this HDD are very interesting: the data transfer rate from the surface in their most dense zone makes 508Mbit/sec (the previous record, 494Mbit/sec, belonged to Barracuda ATA III).

Unfortunately, the relatively complex construction of this drive suffers from increased noisiness (3.7 Bels), although on the other hand, high noise level is not a drawback for a SCSI HDD.

Brief Technical Specs

It would be unfair to compare the performance of Seagate Barracuda 180 with SCSI HDDs rotating faster that is why we decided to include the two hard disk drives with 7200rpm rotation speed, which we have already tested some time ago. They are Seagate Barracuda 18XL and Quantum Atlas V (for more details on these HDDs please see our Seagate Cheetah 36XL HDD Review).

Now let's take a look at the most important HDD features:

As we can see, besides an awful lot of platters and heads, the parameters of Barracuda 180 are quite OK. Average Seek is a bit higher than by other testing participants, however, we hope it won't tell too much on the performance.

Testbed and Methods

Our test system was configured as follows:

• Intel Coppermine 600MHz CPU;
• ASUS CUBX-E mainboard with 1007A BIOS;
• 2 x 128MB PC133 SDRAM by Hyundai;
• Matrox Millennium 4MB graphics card;
• Adaptec ASC26160N SCSI controller;
• Windows 98/Windows 2000 Pro.

To test the HDDs we used Adaptec ASC26160N controller and the ordinary LVD cable. All the drives were tested in Ultra160 mode. For Adaptec controller we took the drivers version 1.00 for Windows2000.

We used FAT32 and NTFS file systems to format each HDD as one logical drive of the maximum size with the default cluster. All the tests were run 4 times and then the average results were taken for the diagrams. The HDDs didn't rest for cooling down between the tests.

Here are the benchmarks used:

• WinBench 99 1.2
• HDTach 2.61
• IOMeter 1999.10.20

We didn't test this HDD under Windows98 because it doesn't make much sense, actually.

Performance

Access Time

Since for our comparison we chose hard drives with 7200rpm rotation speed, we didn't expect to see any grave difference in the results obtained.

Just as we had expected, the access time shown by Barracuda 180 appeared a bit greater than by other HDDs. It can be easily explained by the fact that Barracuda 180 features much more cylinders (24247) than Barracuda 18XL (14384). Moreover, the heads block of Barracuda 180 (which comprises 24 heads, as we have said above) weighs much more and hence is harder to move from one position to another, because it is more complicated to start moving it and then to stop in the needed position. So, we consider the results shown by Barracuda 180 to be very good (we mean that taking into account all the factors, it could have been much worse).

Linear Read Speed

To measure this parameter we used Disk Transfer Rate test from ZD WinBench 99 benchmark set. Among the three HDDs tested, Barracuda 180 can boast the most dense platters: the capacity of each of them makes 15GB, which directly tells on the linear read speed, as you remember.

And undoubtedly this advantage resulted into the highest Disk Transfer Rate (Beginning) value among the tested drives: 42800KB/sec.

Well, finally we see that SCSI drives started getting high capacity platters catching up with today's IDE drives in this term. If the linear read speed of SCSI drives used to increase only due to increased spindle rotation speed, now it's high time the data density got higher as well. As is known, the platters can be decreased in size for the sake of reduced track-to-track seek time. However, in fact, it makes sense for inner tracks only, since their data density is lower. It will not work for hard drives with high rotation speed because the weight of spindle and platters as well as spindle centering matter a lot for them. Therefore, they can get platters of smaller diameter, i.e. their outer tracks get shorter. However, for a HDD with 7200rpm the platter diameter can be cut down in the inner part of the platter, which is less dense. This way we will get a relatively good average seek time and high average read/write speed.

But the linear read graph appeared of a little bit unexpected shape:

As you can see yourself, the nearly ideal linear read graph is disturbed by very grave drops, which usually fall upon the area where two zones with different data density meet. We believe that it was caused by incorrect operation of caching algorithms. Although if we try to imagine how the sectors are physically located on 24 sides of 12 platters, it will hardly be a trivial task. That is why it may be exactly because of this complexity that delays like that occur, when the drive shifts from the zone with one density to that with a different one.

HDTach 2.61

At first, please see the results obtained in this test:

Graphs	Graph	Graph	Graph

And now let's find out how greatly the multi-platter design of Barracuda 180 told on the average read and write speeds:

The average read speed shown by Barracuda 180 appeared 1.5 times higher than that of Barracuda 18XL, however, did you expect to see anything different? Barracuda 18XL features 6GB platters, while Barracuda 180 - 15GB ones.

When writing, the data density isn't of the same importance any more and Barracuda 180 turns out 22% faster than the rivals!

WinBench 99 1.2 FAT32

To tell the truth, we don't think that many of you could afford to use a HDD like that for ordinary work in Windows applications, however, just for the sake of experiment we checked these results as well.

Frankly speaking, we encountered some problems when formatting Barracuda 180 in FAT32 for Win2000. The operation system simply couldn't format a drive that big. So, we had to get the good old fdisk&format and to start formatting. By morning it was all done.

Graphs	Graph	Graph	Graph

Well, the results can't be called brilliant but are quite good anyway. We can see a really noticeable performance increase in some applications (especially for MicroStation SE). Just to make it easier to compare, we composed a diagram for Business Disk WinMark and High-End Disk WinMark results only:

Barracuda 180 performed a bit slower in Business Disk WinMark than Quantum Atlas V, however, in High-End test this HDD became an indisputable leader.

WinBench 99 1.2 NTFS

And as for NTFS, everything ran smoothly without any problems at all. This file system feels to be much better integrated into Windows2000 (is it WinNT's heredity that tells?). But we have to admit that 4KB cluster and a 180GB HDD make a really cool combination :-)

In NTFS Barracuda 180 won in both tests, though with a very small advantage over the competitors. No doubt, that its performance is limited by the cluster size.

Intel IOMeter

Since Barracuda 180 is not an ordinary HDD, we didn't restrict ourselves to the common set of IOMeter benchmarks but added two more patters: RandomWrite (in 8KB portions) and SequentialWrite (in 256KB portion).

We have already described FileServer, WorkStation and DataBase patterns in our previous reviews. Now we would like to add the info for two more patterns for your reference:

Here are the results obtained:

After the first review, where we used this test, we received some feedback where people asked "what's the use of the test, which can transfer only 1.5MB/sec?" The diagram below shows that this benchmark is very flexible and by selecting this or that pattern we can take a closer look at different operation modes.

See yourself: in SequentialWrite the HDD writes 39MB/sec! In case of all other patters, where the sector address is obtained with the help of random numbers generator and the data portions are quite small, we see a sharp drop of the data stream size processed by the HDD per second.

The diagram below shows the dependence of Total I/O on the load:

Well, this is a very illustrative diagram. The graphs for Barracuda 180 in FileServer, WorkStation and DataBase patterns are located very close to one another. Note that the higher is the read percent of the pattern (out of the total requests number), the more requests it can process. With the increase in the number of submitted requests, the difference between the HDD performance for various patters gets smaller. We think it is most likely to be connected with the following. The HDD firmware gets the possibility to queue the requests in the most optimal way and the more requests are submitted to the drive, the more optimal route for the heads can be worked out. Note that the graphs become linear for all the three patterns in the interval between 4 and 16.

The graph for RandomWrite pattern behaves in a pretty typical way: up to the load of 16 requests the graph almost coincides with the graph for SequentialWrite pattern, i.e. its performance is limited by the spindle rotation speed. And the higher gets the load, the greater increases the performance and the results for this pattern catch up with those for FileServer, WorkStation and DataBase patterns. In other words, it means that the heads manage to process several requests addressed to different cylinders during a single turn.

And now let's compare Seagate Barracuda 180 with Seagate Barracuda 18XL and Quantum Atlas V from the Sum Total point of view. The Sum Total is calculated taking into consideration the results obtained with Light, Moderate and Heavy loads (which creates the severest conditions for the drives):

Hm… Barracuda 180 lost everywhere, in all three patterns. Well, and what did you expect from this monster? Large platters, unfortunately, imply more cylinders, which negatively tells on the HDD performance when working with random requests. Although if we recall the results obtained for IDE HDDs with 7200rpm rotation speed, then Barracuda 180 turns out much faster (almost 1.5 times). First of all, if we have a multi-platter drive it is more probable that the needed sector appears on the same cylinder as the previously requested one (the heads won't need to make a step forward). And secondly, this drive features a larger cache-buffer.

Heat and Noise

We measured the heating with the external thermal diode connected to the mainboard. The HDD cooler was disconnected and there were only two coolers blowing the warm air off the case: one in the power supply unit and an additional one on the case rear panel.

After we ran IOMeter for 4 hours the hard drive got only 42 degrees Centigrade warm, which is not bad at all for a 12-platter SCSI device.

As for noise, here is a lot we could say about it. Barracuda 180 HDD is very loud, it is the loudest drive of all we have ever heard working. However, since it is very unlikely to be in your home PC, you will hardly care a lot about how much noise it produces. It is destined to be locked in an empty server room…

Conclusion

It is for the first time that we got hold of a HDD of this type that is why we may have turned out not quite ready to test it properly. First of all, we didn't have the right methodology for testing storage devices of the kind, secondly, we simply didn't have anything to compare Barracuda 180 with. The thing is that there will hardly appear any one extraordinary enough to use this drive in a single HDD configuration. It is too big for a home PC, and those guys who work with video editing may be unhappy about the "drops" we discovered on the read graphs and a not very stable linear write graph. Moreover, all hard disk drives are mortal that is why losing 180GB of data is a considerably greater loss than 18GB, for instance. So, it inevitably comes to mind that this storage device should be used in RAID systems, which automatically throws it into a totally different market segment.

Well, if just for a second we disregarded the HDD's unique construction with 12 platters and 24 heads, its enormous size of 180GB, it would become evident that it is an ordinary SCSI drive for large data storage systems. It is just an ordinary brick, but a very big one :-)

With the help of this solution one can very quickly and efficiently increase the server disk subsystems size. But it is suitable only for those servers, which do not hunt for the shortest reaction to requests. From this viewpoint, Barracuda 180 deserves all the praise.

At present Seagate Barracuda 180 the most optimal solution taking into account the cost of storing 1GB of data and disk subsystems built on it allow very easy scaling. Well, we'll see if it is a really acute topic…