General Framework for Optimization of Distributed Queries

The paper suggests the heuristic approach for selecting the optimal evaluation plan and Semi-join approach for reducing the communication cost. Calculating the cost of each evaluation plan of a query takes lots of computational efforts as well as time. Initially the heuristic approach is used to find best evaluation plan among various plans of a single query. By considering only joins, queries are divided into two: tree and cyclic. Semi-join is a useful tool to reduce the cost of joins. The communication cost in distributed database is mainly depending upon the amount of data transferred from one site to another site. Semi-join approach is sufficient to fully reduce the tree queries while semi-joins are not the full reducers for cyclic queries. Cyclic queries are therefore converted into tree queries using different approaches such as addition of vertices, elimination of vertices, decomposition of vertices and elimination of edges. After conversion of cyclic to tree query, semi-join approach can be used.