A DBMS For Large Statistical Databases

This paper describes the approach taken at Statistics Canada to create a generalized DBMS appropriate for the management of large statistical data-bases. The authors describe the requirements for statistical database processing which differ from more traditional database applications and the methods employed in storage organization and system architecture to satisfy them. Special emphasis is given to the usefulness of the transposed physical structure for the creation of statistical databases. Finally, the applicability of the system as a basis for a relational DBMS for large scale processing is discussed.