Access Methods

We are witnessing a tremendous growth in the size of the data gathered, stored, and processed by various kinds of information systems. Therefore, no matter how big memories become, there is always the need to store data in secondary or even tertiary storage to facilitate access. Even if the data set can fit in main memory, there is still a need to organize data to enable efficient processing. In this chapter, we discuss the most important issues related to the design of efficient access methods (i.e., indexing schemes), which are the fundamental tools in database systems for efficient query processing. For the rest of the discussion, we are going to use the terms access method and index interchangeably. Take, for example, a large data set containing information about millions of astronomical objects (e.g., stars, planets, comets). An astronomer may require some information out of this data set. Therefore, the most natural way to proceed is to store the data set in a database management system (DBMS) in order to enjoy SQL-like query formulation. For example, a possible query in natural language is: “show me all stars which are at most 1000 light-years away from the sun.” To answer such a query efficiently, one should avoid the exhaustive examination of the whole data set. Otherwise, the execution of each query will occupy the system for a long period of time, which is not practical and leads to performance degradation. For the rest of the discussion, we are mainly interested in disk-based access methods, where the data set as well as the auxiliary data structures to facilitate access reside on magnetic disks. The challenge in this case is to perform as few disk accesses as possible, because each random access to the disk (i.e., reading or writing a block) costs about 5–8 ms, which is significantly slower than processing in main memory. Moreover, we assume that our data are represented by records of the form , 10