Fundamentals of Data Base Systems

Essentially a data base is a mass storage of data. Its concept is not a sudden breakthrough in computer technology, rather it is brought about gradually through experience and requirements. Many organisations contributed to the development of data bases; the list is long and includes computer manufacturers, software houses, user organisations and professional bodies. In this chapter we shall first briefly describe the role of data bases, tracing the history of their development. Later sections will be used to clarify some basic concepts and to discuss the advantages and disadvantages of a data base.

As a collection of data, a data base uses the concepts of data items, records and files in one form or another, and it employs the techniques available for their representations in a computer environment. The knowledge of these concepts and techniques therefore forms a prerequisite for a proper understanding of data base technology. This chapter is intended to satisfy this prerequisite. In section 2.1 the characteristics of data items are examined. A general discussion on records and files is given in section 2.2. Indexed files are described in section 2.3 and direct files, along with various hashing techniques are given in section 2.4. Inverted files are presented in section 2.5, with list-structured files and techniques for pointer organisation in section 2.6.

Within the context of an organisation, data are related to one another in a rather complex manner. At any given time, an individual user is concerned with only a subset of these relationships, but a data base, wishing to serve all users, must represent all their facets. This can only be achieved by using complex data structures. In fact the performance of a data base is critically dependent on the data structures it can support and the flexibility such structures can provide. This is why data organisation is such a key issue in data base design. In this chapter we shall discuss the basic problems of data base organisation and examine their solutions.

In the last chapter we discussed data base organisation, which forms the foundation of a data base. Certain key features, such as data independence and the flexibility of data structures, follow directly from this foundation, but there are other features that must be built on the foundation to transform it into a viable data base capable of offering the full range of the expected advantages. These features provide the following facilities. (1) Performance optimisation: facilities to evaluate performance and tune the data base from time to time to optimize its performance. (2) Concurrent usage: the use of the data base by more than one application program at the same time. (3) Data protection: the protection against loss or damage of data in the data base and the protection of the confidentiality of the data from unauthorised persons.

The Codasyl model is a proposed specification for a Data Base System resulting from international collaboration among all interested parties, including users and manufacturers. The specification is designed to fulfil user needs, and as such it incorporates many of the user facilities discussed in the previous two chapters. As mentioned in chapter 1, the model enjoys the active support of most manufacturers, software houses, user organisations and professional organisations such as BCS, ACM and IFIP. A number of data base packages founded on the Codasyl model are already available on the market.

Many Data Base Systems are available today, some being developed by the manufacturers, others by the software houses and other interested parties. The manufacturers’ products are naturally geared to their own machines; those produced by others are usually suitable for a variety of machines. Apart from this machine dependence, there is also a wide variation in the facilities these products offer. Some of these products are better than others in some respects, while worse in other respects; some are complex while others are simple, and some appeal to the larger users while others are suitable for smaller users. Despite this diversity, there is some cohesion with respect to the basic data structures employed and, from this structural consideration, we can divide the products into the following four major groups.

The introduction of a data base as the central reservoir of data affects the user organisation in a number of ways. For example, it changes the organisation’s attitude to data requirements and management, it creates new authorities and it brings in new skills. It also enforces greater coordination between the various user departments and demands stricter adherence to standards. A good implementation scheme should include adequate provision to tackle these problems, in addition to having plans for system developments and scheduling of resources. Much of this would be planned and controlled by the DBA, on whose ability will largely depend the success of the venture—provided that the right DBS is selected in the first place. In this chapter we shall discuss these issues, starting with a general discussion in section 8.1 followed by the role of the DBA in section 8.2, the data base selection criteria being examined in section 8.3. To round off the discussion we shall present some user experience in section 8.4.

Data base technology is barely 10 years old and is still in the early stage of development, but is has already made a great impact on data processing. A growing number of people are now using data bases and there is no doubt that this trend will continue, most probably with an accelerated pace due to decreasing hardware and increasing software costs. However, data bases are slow in processing and consequently costly in operation. The breakthrough here is likely to come from the current researches in storage technology, particularly in associative memory. In fact the widespread use of the relational model, with all its powerful facilities, crucially depends on such a breakthrough. But is the relational model capable of meeting all our future data base requirements? To the academic, the answer is a resounding ‘yes’. And yet to researchers in IBM, where tie relational model was conceived and is still being nourished, the answer appears less clear cut. One of their team led by M. E. Senko has already proposed another model called Data Independent Accessing Model (DIAM). This is a serious proposal and certainly deserves careful examination. Future development will also be affected by standardisation in data bases where work has already begun. In this chapter we shall consider these issues, beginning in section 9.1 with an examination of future storage development, followed in section 9.2 with a look at the DIAM model. The progress in standardisation will be discussed in section 9.3, with a conclusion in section 9.4.