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 ]
The hard disk platters are spinning around at high speed, and the spin speed is not synchronized to the process that moves the read/write heads to the correct cylinder on a random access on the hard disk. Therefore, at the time that the heads arrive at the correct cylinder, the actual sector that is needed may be anywhere. After the actuator assembly has completed its seek to the correct track, the drive must wait for the correct sector to come around to where the read/write heads are located. This time is called latency. Latency is directly related to the spindle speed of the drive and such is influenced solely by the drive's spindle characteristics. This operation page discussing spindle speeds also contains information relevant to latency.
Conceptually, latency is rather simple to understand; it is also easy to calculate. The faster the disk is spinning, the quicker the correct sector will rotate under the heads, and the lower latency will be. Sometimes the sector will be at just the right spot when the seek is completed, and the latency for that access will be close to zero. Sometimes the needed sector will have just passed the head and in this "worst case", a full rotation will be needed before the sector can be read. On average, latency will be half the time it takes for a full rotation of the disk. This table shows the latency for the most common hard disk spindle speeds:
The "average" value is almost always the one provided as a specification for the drive; sometimes the "worst case" number is also mentioned. Sometimes latency is not even mentioned specifically at all, but it can always be calculated using this formula:
Which factors down to this much simpler formula:
The result is a value in milliseconds.
In looking at the table above, notice that the first increases in spindle speed yielded the greatest percentage improvements in performance. As speeds continue to increase, there are diminishing returns for the extra RPMs. Going from 5,400 RPM to 7,200 RPM shaved 1.4 milliseconds off the average latency, but going from 7,200 to 10,000 (which is a bigger jump in both absolute and percentage terms) only reduces it 1.2 milliseconds. At some point companies will likely "max out" on spindle speeds because there won't be any point in increasing further, especially considering the problems that are created when speeds are increased. The 12,000 speed introduced by the Hitachi Pegasus, while very fast, never really caught on as an industry standard. It looks like 15,000 RPM will be the next standard spindle speed for top-of-the-line SCSI drives. It has yet to be seen what price will be paid for jumping to such a high spindle speed; the improvement in latency over standard 10,000 RPM drives is "only" 1.0 milliseconds.As with seek time, figures in milliseconds are big numbers when dealing with computer system performance, but to shave another 1.0 ms off latency from 15,000 RPM would require going to 30,000 RPM, which would be a very significant engineering challenge probably not justified by shaving 1.0 ms off the total access time for the drive.
Again, as with seek times, latency is most relevant only to certain types of accesses. For multiple, frequent reads of random sectors on the disk, it is an important performance-limiting factor. For reading large continuous blocks of data, latency is a relatively minor factor because it will only happen while waiting to read the first sector of a file. The use of cylinder and head skewing on modern drives is intentionally designed to reduce latency considerations when switching between consecutive heads or cylinders on long sequential reads or writes.
Next: Access Time