What is Rotational Delay? (Explained)

What is Rotational Delay?

A rotational delay signifies the time interval between two information requests. In simple words, it is the measurement of the time taken by the rotating hard drive to transfer data and is usually represented in milliseconds (ms).

Technically, it refers to the time taken by the hard drive read and write heads to move from one sector on the disk to another.

KEY TAKEAWAYS

• A rotational delay of a hard disk drive in a computer refers to the amount of time taken by the disks in it to rotate so that the read/write head reaches the desired location of data on them.
• The rotational delay can be cut down significantly if data requests belong to the nearby sectors on the disk. You can also reduce this delay significantly by replicating data on the disks.
• It is easy to calculate the rotational delay of a hard disk drive by dividing the number of revolutions per minute by 60 seconds. The average rotational delay will be just half of this number.
• Rotational delay of a hard disk normally depends on the speed of the discs or its RPM. The higher the RPM or Revolutions Per Minute, the lower will be the rotational latency of the hard disk drive.
• Typically, the rotational delay will be low overall if the read operations are conducted more frequently on the disk than the write operations, instead of being equal or the other way around.

Understanding Rotational Delay

In simple terms, the rotational latency of a hard drive in a computer is the time taken for a particular block of data on its disks to rotate around to the read/write head.

With only that much said about rotational delay, it may seem to be a very simple thing to you.

However, on the contrary, it is not the case in reality. The rotational delay of the hard disk drive typically causes some significant effects in the following areas in particular:

• The performance of the hard disk drive itself in reading and writing data from and onto it
• The overall performance of the computer system

You may know that the platters on the hard disks rotate at a speed ranging anywhere from 5400 revolutions per minute to 15000 revolutions per minute.

And, knowing the fact that the read/write heads are designed to stay at about 3 nanometers off these platters, you may quite naturally think that the heads are hardly able to read the data stored on the minute magnetic fields on the disks.

This is where disk latency comes into play, which helps the read/write head with all that and return data to the computer system.

Ideally, there are three specific factors that determine the disk latency and these factors are calculated to figure out this latency. These factors are:

• Rotational Latency
• Seek Time
• Transfer Time

Though all of these three factors are important for the hard drive to perform well, it is the rotational delay that plays a slightly more important role, for the reasons explained below.

As you may know, data is stored on the disks or the platters coated with magnetic material.

Since it spins, it is not possible for the read and write heads to be placed on all of the data stored on the disks at all times and at the same time.

This is the reason that the disk platters have to spin really fast so that there are no lags in the reading or writing operations.

This helps the desired data to get under the read/write head as and when required.

And it is the time taken by the platter to place the desired data under the read and write heads that is referred to as the rotational delay of the disk.

Typically, in a hard disc drive, the default rotational delay is zero. This is because on modern drives the on-disk cache makes such delay-based calculations ineffective.

How Do You Calculate Rotational Delay?

Ideally, you can calculate rotational delay by the formula: Rotational Delay = Number of revolutions per minute / 60 seconds.

The steps to follow to calculate the rotational delay of a disk with a speed of 10000 RPM, for example, as well as the Input Output Operations Per Second (IOPS) are as follows:

• Divide 10000 RPM by 60 seconds: 10000/60 = 166 RPS
• Divide 1 by the RPS to convert it to decimal: 1/166 = 0.006 seconds per rotation
• Multiply the seconds per rotation by 1000 milliseconds or 6 ms per rotation
• Divide the result in half since rotational delay is considered half a revolution of a disk: 6/2 = 3 ms
• Add with it an average of 3 ms for seek time: 3 ms + 3 ms = 6 ms
• Add another 2 ms for latency for internal transfer: 6 ms + 2 ms = 8 ms

To get the IOPS, divide 1000 ms by 8 ms per I/O: 1000/8 = 125 IOPS

What is the Maximum Rotational Delay?

Typically, the rotational delay is the time taken by the disks to make a full revolution, usually expressed in seconds. This means that the maximum rotational delay may vary according to the Revolutions Per Minute or RPM of the platters of the hard disk drives.

For example, if the disk platters spin at 5400 RPM, the maximum rotational delay or the time taken to complete a single rotation will be:

(1/5400) x 60 = 0.011 seconds.

In this case, the average rotational delay will be just half of it, that is 0.0055 seconds.

Similarly, if the disks spin at 10000 RPM, the maximum rotational delay will be 0.006 seconds or 6 ms, and the average rotational delay will be 3 ms.

What Can You Do to Reduce the Rotational Delay?

The best you can do to reduce the rotational delay of the hard disc is to choose a copy of the data that is closest to the read and write heads, rotationally.

Ideally, rotational delay starts dominating the cost of disk access when the average seek distance is reduced. This can be a serious concern, and to address it, you can duplicate the data at several rotational positions.

You may also choose the duplicate data that is nearest to the read/write head of the disk. This will reduce the rotational delay by reducing the seek time.

However, when you replicate data to reduce rotational delay, the seek distance will be increased as a result of it, because the data will be pushed farther apart.

Also, unfortunately, reducing the effective length of a track and increasing the frequency of track switches at the same time will reduce the bandwidth of a large I/O.

Placing replicas of data on different tracks within the cylinder of a single disk or on other disks will surely help in avoiding unnecessary track switches.

In addition to that, the possibilities of unnecessary degradation of the large and sequential I/Os crossing the boundaries of the tracks should also be eliminated by rearranging the track skews.

Assuming that you are confused enough at this point, some mathematical explanations may be necessary for your better understanding.

Considering R to be the time taken to complete a single rotation on a single disk, the average rotational delay, R_r(1) will be just half of a complete rotation. Mathematically, this means:

R_r(1) = R/2.

Now, if you duplicate the data D times and spread them evenly on the whole track, the distance between each of them will be 360/D degrees. Based on this principle, the average read rotational delay of the disk R_r will be:

R_r(D) = R/2D

On the other hand, if the replicas are placed randomly on the same track, the average read rotational delay will be:

R_r = R/(D+1)

This is typically less beneficial than the other technique.

However, when you have several replicas on the same track, it will increase the average rotational delay, say R_w for writing all the replicas, to:

R_w(D) = R – (R/2D)

Here, you can see that R_r(D) + R_w(D) = R. This indicates three specific situations as follows:

• The overall rotational delay will be reduced if read operations happen more frequently than the write operations.
• If the frequency of read and write operations is equal, changing D will not have any significant effect on the overall rotational delay.
• If the write operations are more frequent than the read operations, then it is best to have no replication.

You must note here that these relationships are not dependent on the value of R. They are true only in the case of the replica propagations in the foreground.

As for the background propagation, replications may be made even when the write frequency is higher than the read frequency.

Therefore, the write costs will be obviously reduced when the nearest copy is written synchronously. It can also be achieved by spreading other copies in the idle periods.

Rotational Delay Vs Seek Time

• Rotational delay indicates the time taken by the read and write heads of the disk to traverse from one sector to another. On the other hand, seek time refers to the time taken by the read and write heads to move from one specific track to the other.
• Rotational delay is not an attribute of most disk scheduling operations because, in most modern systems there is no physical location of the blocks actually available. On the other hand, seek time is used in most disk scheduling operations.
• The rotational delay can be cut down if the following requests belong to the adjoining sector. On the other hand, the seek time can be lowered significantly if the following request belongs to a track nearby or to the same one.
• The rotational delay can be calculated by dividing the number of revolutions per minute by 60 seconds. On the other hand, seek time can be calculated by multiplying the time taken by the read and write heads to cross one cylinder or track by the number of such cylinders or tracks crossed.
• The rotational delay of a disk typically depends on the speed of the spindle of the hard drive. On the other hand, the seek time depends on two specific aspects, such as the data transfer time and the access time.

Conclusion

So, now that you know about the rotational delay in the hard disks, you may very well figure out how important a role it plays in the overall performance of your computer system in terms of reading and writing data.

It is better for a system to have a lower rotational delay, which typically depends on the RPM of the disks.