A. Disk access time is the interval between the time a computer makes a request for transfer of data from a disk system to primary storage and the time this operation completes. To access data stored on a disk, we need to specify the disk address of the desired data in terms of surface/head number, track/cylinder number, and sector number. A disk stores information always from the beginning of a sector and reads information always from the beginning of a track. Hence, disk access time depends on following three parameters:
Seek time When a disk unit receive a read/write command, it first positions the read/write heads on the specified track (cylinder) number by moving the access arms assembly in proper direction. The time required to position the read/write heads on the specified track/cylinder is called seek time. Seek time varies depending on the position of access arms aseembly when a read/write command is recived. If access arms assembly is on the outermost track and the specified track is the innermost one, seek time will be maximum. It will be zero, if the access arms assembly already happens to be on the specified track. Average seek time is thus specified most systems. It is of the order of 10 to 100 milliseconds.
To reduce seek time, some disk systems use multiple read/-write heads on each access arm. For example, a disk system having two sets of read/write heads for each surface (one for inner tracks, and another for outer tracks), will reduce average seek time by half. This is because each read/write head needs to cover and more across only half of the total number of tracks.
Latency. Once disk drive positions read/write heads on specified track, it activates the head on the specified surface. Since the disk is continuously rotating, this head should wait for the specified sector to come under it. This rotational waitting time,i.e required to spin the sector under the head is called latency, also known as rational time, is a variable that depends on the distance of the specified sector from initial position of the head on the specified track. It also depends on rotational speed of the disk that is anywhere from 300 rpm (rotations per minute) to 7200 rpm. Modern ultra fast disks can reach 10,000 to 15,000 rpm or more. Thus, disk systems normally specify an average latency, which is of the order of 5 to 80 milliseconds. Note that average latency of a disk system isequal to half the time taken by the disk to rotate once. Hence, average latency of a disk system with rotational speed of 3600 rpm will be 0.5/3600 minutes=8.3 milliseconds.
Transfer rate. Transfer rate is the rate at which a computer reads/writes data from/to a disk into memory. Once the disk drive positions the read/write head on the specified sector, it reads/write data at a speed determined by the rotational speed of the disk. If rotational speed of a disk is 3600 rpm and the disk has 125 sector/track with 512 bytes/sector, the amount of data transferred in one full revolution of the disk is 125×512=64,000 bytes=64K bytes (approximately). Hence, transfer rate of the disk system is 64,000×3600/60 bytes/second=38,40,000 bytes/second=3.8 Megabytes/second (approximately). Notice that transfer rate of a disk system depends on density of stored data rotational speed of the disk.
Since data transfer time is negligible (due to high transfer rate) as compared to seek time and latency, average access time for a disk system is its average seek time plus its average latency. Average access time varies greatly from one type of disk system to another and is anywhere from 10 to 600 milliseconds.
Since access time for a piece of data stored on a disk depends on its physical location, it is more correct to refer to a disk system as direct access storage device instead of random access storage device. Random access refers to a storage is random access storage. However, we do not observe this distinction strictly and refer to disk system as random access storage devices.