Learn about the technologies behind the Internet with The TCP/IP Guide!|
NOTE: Using robot software to mass-download the site degrades the server and is prohibited. See here for more.
Find The PC Guide helpful? Please consider a donation to The PC Guide Tip Jar. Visa/MC/Paypal accepted.
|View over 750 of my fine art photos any time for free at DesktopScenes.com!|
[ The PC Guide | Systems and Components Reference Guide | Hard Disk Drives | Hard Disk Performance, Quality and Reliability | Hard Disk Performance | Hard Disk Performance Specifications | Positioning Performance Specifications ]
Access time is the metric that represents the composite of all the other specifications reflecting random performance positioning in the hard disk. As such, it is the best figure for assessing overall positioning performance, and you'd expect it to be the specification most used by hard disk manufacturers and enthusiasts alike. Depending on your level of cynicism then, you will either be very surprised, or not surprised much at all, to learn that it is rarely even discussed. :^) Ironically, in the world of CD-ROMs and other optical storage it is the figure that is universally used for comparing positioning speed. I am really not sure why this discrepancy exists.
Perhaps the problem is that access time is really a derived figure, comprised of the other positioning performance specifications. The most common definition is:
Unfortunately, this definition is not universal, and is made complicated by the fact that manufacturers refuse to standardize on even what access time's subcomponents mean. Some companies incorporate settle time into seek time, some don't, for example. And to make matters worse, some companies use the term "access time" to mean "seek time"! They really are not the same thing at all.
In the end though, when you are looking at the ability of a drive to randomly position, access time is the number you want to look at. Since command overhead and settle time are both relatively small and relatively similar between drives, that leaves the sum of seek time and latency as the defining characteristic between drives. Seek time and latency are a result of very different drive performance factors--seek time being primarily a matter of the actuator and latency the spindle motor--resulting in the possibility of some drives being better in one area and worse in another. In practice, high-end drives with faster spindles usually have better seek times as well since these drives are targeted to a performance-sensitive market that wouldn't buy a drive with slow seek time.
Let's compare a high-end, mainstream IDE/ATA drive, the Maxtor DiamondMax Plus 40, to a high-end, mainstream SCSI drive, the IBM Ultrastar 72ZX. (When I say "high end" I mean that the drives are good performers, but neither drive is the fastest in its interface class at the time I write this.) The Maxtor is a 7200 RPM drive with a seek time spec of "< 9.0 ms", which to me means 9 ms. Its sum of its seek time and latency is about 13.2 ms. The IBM is a 10,000 RPM drive with a seek time spec of 5.3 ms. It's sum of seek time and latency is about 8.3 ms. This difference of 5 ms represents an enormous performance difference between these two drives, one that would be readily apparent to any serious user of the two drives.
As you can see, the Cheetah beats the DiamondMax on both scores, seek time and latency. When comparing drives of a given class, say, IDE/ATA 7200 RPM drives, they will all have the same latency, which means, of course that the only number to differentiate them is seek time. Comparing the Maxtor above to say, the Seagate Barracuda ATA II with its 8.2 ms seek time shows a difference of 0.8 ms, or around 10%. But the proper comparison includes the other components of access time. So the theoretical access time of the Maxtor drive is about 13.7 ms (including 0.5 ms for command overhead) and that of the Seagate Barracuda drive 12.9. The difference now is about 6%. Is that significant? Only you can judge, but you also have to remember that even access time is only one portion of the overall performance picture.
Remember that access time is an average figure, comprised of other averages. In fact, access time on any particular read or write can vary greatly. For an illustration, let's consider the IBM 34GXP drive, look at its minimums and maximums, and see how they translate into access time minimums and maximums:
As you can see, there's quite a range! In the real world these extremes will rarely occur, and over time will be "averaged out" anyway, which is the reason that average figures are used. However, it's important to remember that this wide range can occur on any given access, and random perturbations can affect benchmarks and other performance tests.