A Molecular Database Systems Technology

We present a very simple formalism based on parameterized types and a rule-based algebra to survey and identify the storage structure and query processing algorithm building blocks of database management systems. We demonstrate building block reusability by showing how different combinations of a few blocks yield the structures and algorithms of three different systems, namely System R (centralized), R* (distributed), and GRACE (database machine). We believe that codifying knowledge of DBMS implementations is an important step toward a technology that assembles DBMSs rapidly and cheaply from libraries of prewritten components.