A pipelined database machine with efficient join operation

Abstract

Database management has become increasingly important in recent years. How to design an efficient database machine becomes an essential topic. This paper presents a database machine for supporting the primitive operations of a relational algebra. A hash‐sort‐merge strategy has been adopted to implement the join operation. A reconfigurable merger and bi‐directional zero‐time sorter have been employed in the database machine for flexibly sorting and merging various sizes of buckets. We compare the performance of our database machine with that of some other database machines. From our results, when the sizes of operand relations are large, our database machine is superior.