The Seagate Cheetah 10K.6 hard disk drive was announced back in the spring (May 6). Following the schedule, it reached our test lab in the fall. Fall is a hot time in computer business (and many other fields). This is the time when bunches of new products get rolled out. We have already met a representative of the new SCSI HDD generation with 10,000rpm spindle rotation speed. It's the new HDD family from Fujitsu. Right now it's time to get acquainted with Seagate Cheetah 10K.6.

Along with the introduction of the new SCSI drives, Seagate decided to ease users' life and change the marking system for its products. Now it's very simple to tell which drive is newer and faster. Have a look to see what I mean.

The marking will now consist of three fields:

{Cheetah} {10K}.{6}

• {Cheetah} - the first field shows which class the HDD belongs to (in this case it belongs to Enterprise).
• {10K} - the second field indicates the spindle rotation speed of the drive (10K - 10,000rpm).
• {6} - the third field tells what generation of drives with the above listed characteristics this one belongs to.

Well, Seagate has every right to be proud of the amount of HDD generations they have produced :).

Now that Cheetah 10K.6 hits the market, there are still three previous generations of 10,000 rpm hard drives from Seagate available. They are no strangers to us: Cheetah 36XL, Cheetah 73LP and Cheetah 36ES. On the one hand, this proves the conservatism of the SCSI HDD market; on the other hand, it suggests that none of the above-mentioned drives met users' demands in all market segments. Cheetah 36XL offers too small storage capacity according to the today's standards, Cheetah 36ES isn't fast enough for "heavy" servers and the storage capacities of Cheetah 73LP were somewhat larger than many users could cope with. By the way, Cheetah 73LP HDDs were first introduced in the 73GB version only, but later, insistent requests from OEM partners made Seagate roll out the two-platter version of Cheetah 73LP with the 36GB capacity.

Simultaneous manufacturing of three SCSI drive families with the same spindle rotation speed hardly complies with the low prime cost requirements, which the crisis dictates to the industry. In current conditions, Seagate would be better off offering a drive that would combine high speed, big capacity, low price and low noise level. And they did make such a drive!

One platter of Cheetah 10K.6 can store the same amount of data as four platters of Cheetah 36XL while its maximum capacity is twice as big as that of Cheetah 73LP. The junior models produce a little over 3bel of noise thus equaling the noise level of some IDE HDDs with lower spindle rotation speed. In this review we are going to check if the last of the above mentioned statements is true. Namely, we will dwell on the speed of the new drive from Seagate.

Testing Participants

The main hero of the today's testing session is Seagate Cheetah 10K.6. We were lucky to get the top model of the family with 146GB storage capacity.

Unfortunately we can't use here the results obtained for Cheetah 36XL and Cheetah 73LP, which we got earlier, as benchmarks patterns and drivers for the Adaptec 29160N controller have changed since then. But we do have the results for Cheetah 36ES and that's what we will compare the new Cheetah 10K.6 with.

I am absolutely aware that these hard drives belong to different weight categories, but so far I don't want to compare the last-generation drives from different manufacturers with one another. Our point is to show the progress in speed of the modern drives from every maker (Fujitsu, Seagate, IBM and Maxtor) and then pit the last-generation drives against each other in the final race.

We will also discuss the performance of Cheetah 10K.6 with the U320 interface. The new Seagate's drive appeared to be getting along with the v.1.0.000.000 driver for Adaptec 39320D quite nicely.

The main features of Seagate SCSI HDDs of the last four generations are listed below:

Well, Seagate Cheetah 10K.6 looks most impressive against the background of the other models: the lowest seek time and 8MB cache-buffer bring it to the spotlight. And if recall its huge capacity…

Testbed and Methods

The hard disk drives were tested with the Ultra160 SCSI and Ultra320 SCSI interfaces. Ultra320 SCSI tests were run on the Adaptec 39320D controller.

For WinBench tests the arrays were formatted in FAT32 and NTFS as one partition with the default cluster size. The WinBench tests were run five times each; the average result was taken for further analysis. The HDDs didn't cool down between the tests.

We used the following benchmarking software:

• WinBench 99 1.2
• HDTach 2.61
• IOMeter 1999.10.20

To compare the hard disk drives performance in Intel IOMeter we used the StorageReview patterns described in the third edition of their HDD testing methodology.

These patterns are intended to measure the disk subsystem performance under workloads typical of file- and web-servers.

Our colleague, Sergey Romanov aka GreY, developed a pattern for Intel IOMeter basing on the StorageReveiw's study of the disk subsystem workload in ordinary Windows applications. The pattern was based on the average IPEAK statistics StorageReview provided for Office, High-End and Bootup work modes.

The pattern serves to determine the attractiveness of the HDDs for an ordinary Windows user.

Well, and in the end we checked the ability of the drives to work with sequential write and read requests of variable size.

Our testbed was configured as follows:

• ASUS P3B-F mainboard;
• Intel Pentium III (Coppermine) 600MHz CPU;
• 2 x 128MB PC100 ECC SDRAM by Hyundai;
• IBM DPTA 372050 HDD;
• Matrox Millennium 4MB graphics card;
• Windows 2000 Pro SP2.

The controllers were used with the following drivers:

• Adaptec 29160N - 4.10.4002;
• Adaptec 39320B - 1.0.000.000.

The reviewed drives had the following firmware versions:

• Cheetah 36ES - 0108;
• Cheetah 10K.6 - 0001.

Performance

HDTach 2.61

The first benchmark to run is HDTach:

The results proved exactly what we had expected them to be. Average seek time is 0.5ms lower by Cheetah 10K.6 than by Cheetah 36ES, and Cheetah 10K.6 outperforms Cheetah 36ES by 4-5MB/sec in the average read/write speed.

Well, we have already seen a hard drive behave like that with U160 and U320 SCSI (even two times! :)). Yes, Cheetah 10K.6 with U320 interface is slower than with Ultra160 when reading from the buffer, but the gap isn't that big as it comes to the average read speed. And there is no gap at all in average write speed…

WinBench99 1.2

Our previous benchmarking experience suggests that WinBench99 is very sensitive to the HDD's cache-buffer capacity. Let's see if the "rule" holds true here…

First come the results in FAT32 file system:

Linear read graph of Seagate Cheetah 10K.6 146GB is available here.

Pay attention to the linear read speed of Cheetah 10K.6 in the beginning of the disk: nearly 69MB/sec!

As we see, Business Disk WinMark results of Cheetah 36ES and Cheetah 10K.6 differ not that much, while in the High-End test the new drive is beyond competition. It's 17% faster than Cheetah 36ES!

Now let's check the results in NTFS:

In NTFS, the Business test shows equal results again, while in the High-End test Cheetah 10K.6 still outperforms Cheetah 36ES, although the gap is somewhat smaller this time: only 10%.

So, the big cache-buffer and (maybe) higher data density made Cheetah 10K.6 run faster in WinBench tests.

Intel IOMeter: Sequential Read & Write

The table below lists the sequential read speeds of the HDD when the size of the requested data block changes. The request queue depth is set to four requests.

As we see, the new Seagate drive is faster than Cheetah 36ES when processing data blocks of over 1KB.

Now, let's check the writing:

In fact, it is really hard to surprise me now, but Cheetah 36ES did make it nevertheless… Watch this drive processing 512B and 1KB data blocks: it competes with Cheetah 10K.6. But then, on coming over to 2KB blocks, the read speed of Cheetah 36ES becomes significantly lower!

Intel IOMeter: File & Web Server

Cheetah 10K.6 greatly outperforms Cheetah 36ES, no doubt. But let's not forget that Cheetah 36ES isn't the fastest drive on earth. Let's compare the average speeds of the drives across each of the patterns. We calculated the arithmetic mean of Total I/O values under every workload:

Both patterns show that Cheetah 10K.6 has definite advantage over Cheetah 36ES.

Intel IOMeter: WorkStation

The HDD ratings in the Workstation pattern are calculated in different way than in FileServer and WebServer patterns. As the workstation is most probably to encounter smaller workloads, the Total I/O values under low workloads will have a bigger weight in the total result. The suggested formula for calculating the total result looks like this:

Performance = 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.

We put the Total I/O values into the formula and created the following diagram:

As the total result formula in WorkStation is completely different from that we used in FileServer and WebServer, the ratio between the results for Cheetah 10K.6 and Cheetah 36ES is different. Cheetah 10K.6 performs better mostly with long request queue depths, and those are put into the total formula here with diminishing coefficients…

Conclusion

Our tests showed that Seagate rolled out a very successful product, combining high speed with huge capacity. This hard drive will nicely fit into a high-performance workstation (thanks to its capacity and low noise level). Bedsides, it is also very suitable for all types of servers: from file-servers to NAS/SAN systems.

The low storage cost per megabyte and high speed of Cheetah 10K.6 will surely help it to replace all other SCSI HDD models from Seagate with 10,000rpm spindle rotation speed and free the assembly lines for…

But that's another story. :)