Evaluation of Scheduling Policies in a Staged-Database System

Traditional database systems execute queries using one query multiple operators approach. Such systems do not cash on the common data or computation which could be used by multiple queries. Thus leading to poor performance. To overcome these deficiencies, Staged-DB approach has been proposed, where the philosophy is one operator multiple queries. A Staged-DB system splits a traditional DBMS into a number of stages and handles a stage independently. In this project we evaluate the performance of a Staged-DB system under different scheduling policies. Our study is limited to scheduling various components that constitute the execution stage of a Staged-DB. We evaluate different scheduling policies that fix the order in which various operators in the execution stage should be executed. We demonstrate the performance of the policies in terms of response time, throughput, and memory consumption. General Terms Database management systems