End user demand for increased hard drive capacity is waning somewhat, now that we can store over 200,000 typical digital photos, or 60,000 music files, for less than a $100 street price. Sixty thousand music files could play for almost six months without repeating and you'd have to stay awake for almost 70 days straight if you look at each of those 200,000 photographs for 30 seconds each. That's not to say the need for increased storage capacity is waning in the enterprise, but given the choice of a faster 250GB hard drive or an average performing 300GB drive for desktop systems, more enthusiasts are likely to go with the relatively large, but faster 250GB drive. While the race for hard drive capacity on the desktop may be slowing to a simmer, the race for higher performance is heating up. New interfaces, higher bus transfer rates, faster rotational speeds and improved drive software are combining to deliver impressive performance.

One of the most recent additions on the software front is Native Command Queuing (NCQ). NCQ attempts to reduce latency and increase throughput by ordering the incoming read and write requests in the most efficient sequence to minimize head travel. For example, let's say a system asks for the data at position 20 on a drive, immediately followed by a request for more data at position 19. A drive without NCQ will read the data at position 20, and then wait for the platters to make nearly a complete rotation to read the data at position 19. The NCQ enabled drive will rearrange the requests for 20 > 19 into 19 > 20 so the drive can read the data more efficiently without the additional rotation.

On the hardware front, we have products like the latest integrated RAID controller offering from Intel on their ICH7 Southbridge. Back in 1998, when the rumblings of IDE RAID first began, most of us were running 430TX based motherboards with a maximum hard drive transfer rate of 33 MB/s. The first practical implementations of IDE RAID, such as the Promise FastTrak66, supported up to 4 drives in RAID 0, 1 and 10 formats and were limited by the bandwidth of the PCI bus. But today, high-end systems have far more available bandwidth and just about all enthusiast class motherboards include some form of integrated RAID functionality.

In this article, we are going to explore the performance of Maxtor's DiamondMax 10 hard drive as it compares to last year's DiamondMax 9, and the current SATA HD performance leader, the Western Digital Raptor WD740GD. We will then combine the drives in various quantities, RAID formats and stripe sizes to measure the performance impacts of these configurations at relatively similar price points. The RAID formats we will be using are:

• RAID 1, which is basic mirroring, writes all data identically to two drives. Write performance is generally slightly lower than a single drive since data is written twice by two drives that may start in different relative positions. Read performance could be much higher since any read operation could be broken in half with each drive doing half the work, but typically, only a single drive performs read operations giving similar read performance to a single drive.

• RAID 0, which is basic striping, takes all data read and write operations and spreads them across all drives equally. This is the highest performance version of RAID, but also increases the risk of data loss since any single drive failure of the array causes the loss of all data in the array.

• RAID 5, which is striping with parity, is similar to RAID 0, except that each piece of data written to a drive has a piece of parity data written to another drive. In this configuration, loss of a single drive will not corrupt the entire array, but the overhead to calculate the parity data can have a significant impact on write performance.

Finally, we will discuss the impact of stripe sizes on RAID performance. The stripe size is set during array initialization. It represents the size of the pieces a file is dividend into for writing across the drives. For example, a 1 MB file written to a RAID 0 array using a 4K stripe size would be broken into 256 pieces that are spread across all the drives in the array. The same file using a 128K stripe size is only broken into 8 pieces. In RAID 5, the data is broken into pieces according to the stripe size, and additional pieces are calculated for parity error checking data.

The Combatants Spindle Speed, Capacity & Price

    
Maxtor's DiamondMax 10

Maxtor's DiamondMax 10 drive boasts some impressive improvements over last years DiamondMax 9 drive, including double the buffer, NCQ support and what Maxtor calls a "SATA-II" interface although the standard is still taking shape. The Raptor uses its high rotational speed to get the required position under the drive heads in a very short time as demonstrated by the low average seek time. That performance comes at a major price premium though, as the Raptors cost almost 6 times the price/GB of the DiamondMax 10 drive. At those prices, some even high-speed SCSI solutions start to look comparable.

The HotHardware Test System Intel Inside

Our test platform was based on an Intel D945GTP motherboard, which has the ICH7 Southbridge and four SATA connectors supporting NCQ and Intel Matrix Storage Solutions RAID levels 0, 1, 10 and 5.

Notes:

1. The operating system used a separate hard drive from the tested configurations.
2. The tested drives were unformatted for HD Tach tests to enable write testing.

Single / Mirrored Performance Performance with HD Tach

As expected, the read performance of the single drives is nearly identical to the read performance of a mirrored pair of the same type. The Raptor is still the fastest at almost 20% faster than the DiamondMax 10 and just over 30% faster than the DiamondMax Plus 9.

The Raptors are slightly faster writing data than reading it. Here the Raptor improves even more over the DiamondMax drives with nearly a 20% improvement on the DiamondMax 10 and a 43% improvement on the DiamondMax Plus 9.

The burst performance story is quite different with Native Command Queuing showing its potential here. The Maxtor NCQ drives are 13% faster than the Raptors and 49% faster than the DiamondMax Plus 9 drives. This indicates that in instances where the NCQ was able to optimize the incoming requests, significant improvements are realized.

Single Drive & Mirrored Performance (Cont.) Performance with HD Tach

The slight increase in CPU utilization between a single drive and a mirrored pair of drives is expected. The 0% utilization for the pair of Raptors in mirrored mode was not expected. But the test was repeated several times with the same results.  Either the Raptors averaged less than 1% CPU utilization or HD Tach wasn't able to record the data accurately in this mode.

The higher rotational speed of the Raptors consistently brings the data to the read heads faster. It is interesting to note as well that the DiamondMax 9 drive is slightly faster getting to its data than the DiamondMax 10.

The performance differences are consistent across the platter surface with the Raptor (blue) on top, DiamondMax 10 (red) next and the DiamondMax Plus 9 (green) slowest.

Single/Mirrored Drive Summary:
Clearly, Maxtor has made some improvements in overall performance on the DiamondMax line between the 9 series and 10 series. Aside from the slightly lower seek time of the DiamondMax Plus 9, the DiamondMax 10 was better in every category. The Raptors pure rotational velocity still gives it the advantage in sustained operations. However, the NCQ capabilities on the DiamondMax 10 are able to overcome the 2800 RPM deficit against the Raptors under the right conditions.

A 32K stripe size is default for several of today's popular RAID configurations and represents a fair middle ground in the available sizes ranging from 4K to 128K. The 32K stripe size is available in all configurations and provides a good baseline to measure RAID performance.

RAID Testing (32K Stripe) Performance with HD Tach

Clearly, the number of drives used is a significant factor in RAID performance. Adding the third drive to the Maxtor stripe set increases capacity by 50% and read performance by 45%. An interesting item to note is that when adding a third drive to the pair of RAID 0 Maxtor drives and the array is configured in RAID 5, both the capacity and performance are nearly identical. A third Raptor was not available at the time of testing to see if the performance/capacity pattern was drive independent.

For write performance in RAID 0 modes, the pattern is consistent with the read performance. Adding the third drive to the RAID 0 set has a more pronounced impact on write performance, and with a 50% increase in capacity comes a 61% increase in performance. The story is very different with the RAID 5 array where write performance is dramatically reduced by the parity calculation overhead. In RAID 0, the three Maxtor drives are about 6% faster in write operations. In RAID 5, the three Maxtor drives are over 80% slower in write operations.

Similar to the pattern seen in the single drive configurations, the NCQ enabled DiamondMax drives really start to shine in burst measurements. When the conditions are right for NCQ to perform, the numbers clearly show the impact. In two drive RAID 0 configurations, the Raptors are 24.9 MB/s faster in sustained read over the pair of Maxtors. For burst performance, the same configuration of Maxtors are 22.6 MB/s faster than the Raptors. In burst performance, the addition of a third drive to the Maxtor RAID 0 set does not have as pronounced of an impact, but still turns in an impressive number of 316.2 MB/s. This is more than twice the theoretical bus transfer rate of a single Maxtor drive.

More RAID Performance (32K Stripe) Performance with HD Tach

In seek time, the rotational velocity of the Raptors still overcomes all configurations of the Maxtor drives. The pattern of a pair of Maxtor drives having very similar performance to three drives in RAID 5 performance is maintained. We would have expected additional read heads on a three drive RAID 0 configuration of Maxtor drives to reduce seek times, but this was not the case.

It is very normal to see high CPU utilization for RAID 5 configurations. This is not the case when the Intel ICH7 integrated RAID is paired with the NCQ Maxtor drives though. This combination yields utilization numbers generally in line with the other configurations.

RAID 32K Stripe Summary:
The number of drives seems to be a key factor in RAID performance with pure striping. In RAID 5, the severe write performance penalty should make most users think twice about this configuration. Certain systems where write requirements are minimal and up-time through a drive failure is critical, like a web server, are good candidates.

Here we compare the two fastest configurations of drives from our earlier tests: the RAID 0 configurations using three Maxtor drives and two Raptor drives. Three different stripe sizes are configured on the two drive setups to show the impact stripe size has on performance.

Impact of Strip Size on Performance Performance with HD Tach

As you can see, in regard to read performance, smaller stripe sizes seem to offer better performance with 17-18 MB/s increased average throughput for the 4k stripe versus the 128k stripe.

The read performance pattern in repeated here. The pattern of write performance is similar, where the smaller stripe sizes offer similar in performance that falls off with the larger stripe sizes.

With burst performance, the stripe size seems to have little impact on the Raptor drives where there is less than 2% difference for all stripe sizes. In the three drive Maxtor arrays, the highest number in any of the testing for this article are seen with the smallest stripe size, with 343.3 MB/s on the 4K stripe size array.

Impact of Stripe Size on Performance (Cont.) Performance with HD Tach

Stripe size seems to have almost no impact on seek time for either configuration.  And once again, the higher rotational speed of the Raptor drives makes them the performance leader.

Here we start to see the price of the smaller stripe size. When a 1MB file is broken in hundreds of pieces rather than dozens, it requires more CPU power to disassemble/reassemble all the parts.

In RAID 5, the opposite pattern is true. The larger stripe sizes result in fewer data slices, which means means fewer parity calculations. In RAID 5, the two larger sizes are almost identical in performance where the smallest stripe size (16K for RAID 5 rather then the 4K available in RAID 0), is slower. In any case, the dismal write performance must be a serious consideration.

Stripe Size Performance Summary:
There is no hard and fast rule about stripe sizes and RAID performance. Additionally, the usage of the array is as critical as the components that make it up. In general, RAID 5 should only be used where system survivability through a drive failure is critical or on systems where write performance is not significant.

The performance improvement offered by NCQ and a faster SATA interface on the Maxtor DiamondMax 10 line of drives is more evolutionary than revolutionary when compared to the DiamondMax 9. And these new features are still not enough for the new Maxtor drives to take the overall performance spot from the Western Digital WD740GD Raptor drives.

However, when pairing the new NCQ enabled Maxtor DiamondMax 10 drives with the Intel ICH7 integrated RAID controller in the right combinations and quantities, some very impressive performance numbers start to emerge. The maximum theoretical combined transfer rate of three Maxtor DiamondMax 10 drives is 450 MB/s. In a RAID 0 configuration with a 4K stripe size, we recorded a burst speed of 343 MB/s, which is about 75% of the theoretical limit.

When configured in RAID 0, the Maxtor drives have faster performance and almost 5 times the capacity of a pair of Raptors. When you factor in the nearly $50 lower price tag on the Maxtor drives, the choice for both performance and capacity is clear.  Overall, we were impressed by the value of Maxtor's DiamondMax 10 drives and would recommend them to anyone looking for a high-performing, high-capacity hard drive.  And if you've got the funds, and a controller capable of RAID, performance can be increased dramatically by running two or more of these drives in a striped array. We're giving the 250GB Maxtor DiamondMax 10 hard drives a solid 9 on the Heat Meter.

Get into HotHardware's PC Hardware Forum Right Now!