Efficient algorithms for distributed query processing in star and extended-star networks

Scheduling algorithms for processing queries in a star or extended-star network are presented. A relational database system is assumed, and the data transmission time for joins of attributes is the dominating factor in the query processing. The first algorithm STAR-RT for minimizing the response time of a star query is based on the principle of dynamic programming. The second one, STAR-CT, is an approximation algorithm also for processing a star query, but it aims at minimization of the total communication time. Two optimization subproblems are derived in connection with the query processing and STAR-CT is based on the strategies for solving these subproblems. The above algorithms are extended to those for more general query processing.<<ETX>>