No abstract

This paper describes the Query Rewrite facility of the Starburst extensible database system, a novel phase of query optimization. We present a suite of rewrite rules used in Starburst to transform queries into equivalent queries for faster execution, and also describe the production rule engine which is used by Starburst to choose and execute these rules. Examples are provided demonstrating that these Query Rewrite transformations lead to query execution time improvements of orders of magnitude, suggesting that Query Rewrite in general—and these rewrite rules in particular—are an essential step in query optimization for modern database systems.

This paper describes the Query Rewrite facility of the Starburst extensible database system, a novel phase of query optimization. We present a suite of rewrite rules used in Starburst to transform queries into equivalent queries for faster execution, and also describe the production rule engine which is used by Starburst to choose and execute these rules. Examples we provided demonstrating that these Query Rewrite transformations lead to query execution time improvements of orders of magnitude, suggesting that Query Rewrite in general -and these rewrite rules in particular -are an essential step in query optimization for modern database systems.

We introduce a class of novel multiprocessor scheduling problems that arise in the optimization of SQL queries for parallel machines. These consist of scheduling a tree of interdependent communicating operators while exploiting both inter-operator and intra-operator parallelism. We develop algorithms for the specific problem of scheduling a Pipelined Operator Tree in which all operators run in parallel using inter-operator parallelism. Weights associated with nodes and edges represent respectively the cost of operators and communication. Communication cost is incurred only if adjacent operators are assigned different processors. The optimization problem is to assign operators to processors so as to minimize the maximum processor load. We develop two approximation algorithms for this NP-hard problem. The faster algorithm has a performance ratio of 3.56 while the slower algorithm has a ratio of 2.87.