Database Processing Models in Parallel Processing Systems

This paper investigates database processing in parallel processing systems. The objective is to maximize throughput and minimize response time through conflict free data accesses. We propose abstract models as a general framework, and then present two specific parallel processing strategies for two common types of database applications. One is optimal file distribution for partial match queries, and the other is a multidirectory hasing scheme where database accesses are based on primary keys. We show that these proposed parallel processing strategies perform better than those proposed earlier. This work presents a new basis on which design of parallel processing database systems for various applications can be facilitated more systematically.