Secondary Key Access Methods

Abstract

Access methods for secondary key retrieval have attracted much interest. The problem is stated as follows: Given a file, say, EMPLOYEE( name, salary, age), organize the appropriate indices so that we can answer efficiently queries on any and all of the available attributes. Rivest [Riv76] classified the possible queries into the following classes, in increasing order of complexity:

• exact match query, when the query specifies all the attribute values of the desired record, e.g.:

  name = ‘Smith’ and salary = 40,000 and age = 45

• partial match query, when only some of the attribute values are specified, e.g.:

  salary = 40,000 and age =35

• range queries, when ranges for some or all of the attributes are specified, e.g.:

  35,000 ≤ salary ≤ 45,000 and age = 45

• Boolean queries:

  ( (not name = ‘Smith’) and salary ≥ 40,000 ) or age ≥ 50