Design of a distributed CODASYL database system

A general model for distributed database systems based on the three-level architecture is presented. How concrete architectural forms can be derived from the general model is illustrated by means of the CODASYL database model. A new language is introduced, the data partition language, in which all distribution specific statements are concentrated. The application of distribution predicates in the data partition language to form distribution units leads to the concept of a “distribution graph”—it is a subset of the schema graph and defines the validity scope of a distribution predicate. Global database keys, global currency indicators and multiple record access are suggested to manipulate distributed CODASYL databases.