Desktop version

Home arrow Computer Science

  • Increase font
  • Decrease font


<<   CONTENTS   >>

Serial-Access Memory: External Memory

Primary memories realized by contemporary semiconductor technologies are quite fast due to the directly accessible and independently addressable (random) features of each of its storage location. But, one of their major limitations is that they are volatile, i.e. permanent storing of information is not possible. Moreover, their storage capability is limited, mainly due to the cost consideration in relation to per bit of stored information, and hence, they cannot fulfil the storage-space requirements of the computer systems. Large storage requirements with permanent storing of most general-purpose computer systems are thus fulfilled with the use of serial-access memories (often called secondary/external memory). The most commonly used memories of this category are economically realized in the form of magnetic disks, magnetic tapes, optical disks, and nowadays flash drives (pen drives). These memories except flash drives (already discussed) will now be discussed in the subsequent sections one after another.

Characteristics

Most computers have slower, cheaper, and usually much larger serial memories to permanently hold bulk volume of data that are usually stored on a set of fixed paths or tracks. Tracks are a concentric set of rings (or a set of parallel lines), and each track consists of a sequence of cells; each cell (storage location) is capable of storing 1 bit of information. Read-write circuitry is here shared among different storage locations. Every track has over it a number of fixed access points at which the information can be transferred to or from the track. Transfer of information to and from any track is accomplished by moving either the stored information, or the read-write unit (Head) or both. Positioning the read-write unit (Head) on the desired track is a direct one, and then the information being transferred to and from the accessed track is essentially serial. The larger size of serial memories realized by electromechanical equipment using magnetizable material to construct small simple storage cells is low-cost (inexpensive) and also portable, but they provide relatively long access time which is mostly due to the following factors:

i. The time taken to position the read-write unit (Head) on the desired track.

ii. The time elapsed due to relatively slow speed of the track movement.

iii. The time taken to transfer the data serially, and not in parallel to and from the memory.

Apart from the size to be accounted, the access time of serial memory is also considered as a critical parameter at the time of its selection. The major characteristics that create the differences among the various types of serial memories are [Мее, C, et al.]:

Head/Storage media movement: In a movable-head system (except a fixed-head system), there is only one read/write head which is mounted on an arm, and the arm, in turn, can be extended or retracted over the tracks.

Access Procedure: In a movable head system, the only read-write head is needed to move between different tracks. The average time required to move a head from one track to another is called the seek time Ts of the memory. Once the head is placed on the desired track, the targeted cell may or may not be just under the head. A certain amount of time is thus needed for the targeted cell to reach under the read-write head so that data transfer can begin. The average time needed for this movement of the cell is called the latency or sometimes called the rotational delay TL of the memory. A substantial amount of time has thus been already consumed by way of seek time and latency time for only to reach the targeted information. So, once the destined word is found, it is not economical and even inefficient to access just one word per access. A chunk of words is, hence, normally accessed in place, and these words are usually grouped into a larger unit called block. An entire block consisting of a group of words is then accessed using only one seek and one latency time instead of many.

Data Transfer Time: To compute an estimated average time required to transfer the entire information of a sector (consisting of several blocks) or a block (consisting of several words) in a serial-access memory, we assume for the sake of simplicity that the memory has closed rotating tracks (disk), and this time then essentially consists of three fundamental components:

a. First the head(s) is(are) to be placed on the right track (cylinder) containing target sector(s)consuming an average seek time, say, Ts. Typical value of Ts is usually about 9 ms.

b. Head is now placed on the track. The start of the desired sector on the track may be just under the head, or may be at any position on the track. Hence, on an average, half the revolution is required to get the beginning of the target sector under the head. This time is the latency time TL, or sometimes called average rotation which is,

c. The read-write head is now positioned at the starting point of the requested sector(block). Transfer of data can now be started. The rate at which the data is transferred depends primarily on:

  • - the speed at which the stored information is moving under the head. This is expressed in terms of revolution per minute (rptn). For example, the rpm of a rotating device is 7,200, it means that the device can accomplish 7,200 rotation per minute. This characteristic is often supplied by the vendor.
  • - the number of bits to be transferred depends on the storage density along the track. The density is defined in bits per unit length.

The rate at which the information is transferred continuously to or from a track under any situation is called the data-transfer rate. If a track has a storage density of k bits per centimeter and moves at a velocity of v centimeter per sec past the head, then the number of bits being transferred per sec is kv. This kv is considered as the data-transfer rate. Suppose each track has an average capacity of N sectors (blocks), and each track is rotating at the speed of r revolutions per second, then; the data-transfer rate of the device is r x N sectors (blocks) per second. So, the time required to transfer 1 sector (block) is = l/(r x N) second. In other words;

Therefore, the total average time required to transfer the entire information from any specified target sector is:

Platters: A single-platter (a single circular platter) constructed of plastic, coated with a magnetizable material forms the disk. Data are recorded on circular tracks already created on the disk, and later retrieved from the disk via the head. During a read/write operation, the head is movable while the platter rotates beneath it. Floppy disks and CDs are of this type. Some disk drives accommodate multiple platters stacked vertically about an inch or less apart. Each one is fixed on a spindle at a definite distance along the length of the spindle. Multiple arms are provided in between the platters. Each arm has two heads; one for the lower surface of the upper platter, and the other is for the upper surface of the lower platter. In fact, each arm moves radially through inter-platter gap. All the arms now move together back and forth in a fixed linear path so that all heads move in unison to select a particular track or a set of tracks. The spindle is given a circular motion with the help of a rotating shaft attached to an electric motor to give a constant circular motion to the attached platter. The entire unit is known as a disk pack.

Sides: If the magnetizable coating is applied to one side of the platter, the disk is known as a single-sided disk. Floppy disks of the early days, and nowadays CDs are of this type. For most disk (hard disks), the magnetizable coating is applied to both sides of the platter and are referred to as double-sided. Today's hard disk is of this type.

Portability: A non-removable disk permanently mounted in the disk drive is a sealed pack, and it is used as one unit. All the hard disks used in today's computers (except mainframes) are non-removable disks. A removable disk is mounted in a disk drive. The disk drive consists of an arm with the read/write head attached to it, a shaft that rotates the disk, and the entire electronic circuits required for transferring (input/output) binary data. The removable disk may be of single platter (CDs) or may be of multiple platter (disk pack used in a mainframe large system). The removable disk can be taken away manually from the disk drive and replaced with another disk on the same drive even when the computer is on. The advantage of this type of disk is that an unlimited amount of data is available with a limited number of disk systems. Furthermore, a disk may be moved from one computer system, and it can be mounted in another compatible system.

Head Mechanisms: It is an important parameter in electromechanically accessed magnetic memories that distinguishes between magnetic-disk and -tape memories, and also provides a clear classification of the disk into three types. The Contact type (magnetic tape) mechanism allows the head to come into physical contact with the surface of the magnetic medium during a read/write operation. In the Fixed-gap type, the read/write head is positioned a fixed distance apart above the magnetic surface, allowing an air gap. This very narrow space separates the head from a cell on the storage track of the surface, so that the action of the head can interact to transfer information between the head and the storage medium. In Aerodynamic-gap type (Winchester), the mechanism is used in a sealed drive assemblies that almost have no contaminants (dust, pollution, etc.). The head here in the shape of an aerodynamic foil rests lightly on the platter's surface (magnetic surface) when the disk is not in motion. When the disk comes into motion, the air pressure generated by the rotating disk is enough to make the foil rise above the surface to operate closer to the disk's surface but not in contact. It can then interact with the platters' surface to transfer information between the head and the storage medium. Nowadays, any aerodynamic head design used in a sealed-unit disk drive uses the term "Winchester".

A brief detail of these topics with the respective figures and a solved problem based on these characteristics is given in the following web site: http://routledge.com/9780367255732.

 
<<   CONTENTS   >>

Related topics