Improved and Optimized Partitioning Techniques in Database Query Processing

. In this paper we present two improvements to the partitioning process: 1) A new dynamic buffer management strategy is employed to increase the average block size of I/O-transfers to temporary files, and 2) An optimal switching between three different variants of the partitioning methods that ensures minimal partitioning cost. The expected performance gain resulting from the new management strategy is about 30% for a reasonable resource configuration. The performance gain decreases with increasing available buffer space. The different partitioning strategies (partial partitioning or hybrid hashing, one pass partitioning, and multipass partitioning) are analyzed, and we present the optimal working range for these, as a function of operand volume and available memory. Keywords: Relational algebra, partitioning methods, buffer management, query processing 1 Introduction Relational algebra operations are time and resource consuming, especially when done on large operand volumes. In this ..