ABSTRACT

The typical organization of a database is as a collection of files, each containing a collection of records, stored on a set of disks. Throughout this chapter we assume disks, either magnetic or optical, are being used

as secondary storage media. We noted the main physical characteristics of a computer disk in Chapter 1. The atomic unit of data held on a disk is referred to as the disk block. The time taken to transfer a disk block to orblock from a disk has three components:

• the time taken for the mechanical movement of the disk heads across the disk to the correct track, termed seek time;

• the time taken for the disk to rotate to the correct position, termed latency; andlatency

• the time required to transfer the block into the CPU, CPU transfer time.CPU transfer time

As noted in Chapter 1, seek time is the dominant factor determining performance in data retrieval from disks. Data structures in secondary storage that lessen the mechanical movement of the disk heads will therefore result in improved database performance. To minimize disk head movement, data is ideally placed in such a way that blocks which are often accessed together are close together on the disk. Lessening disk head movement will also be a consequence of the construction of appropriate indexes, so that unnecessary searches, maybe of entire files, are avoided.