Query Processing and Optimization in Distributed Database Systems

Summary 

Query processing is an important concern in the field of distributed databases. The main problem is: if a query can be 

decomposed into subqueries that require operations at geographically separated databases, determine the sequence and 

the sites for performing this set of operations such that the operating cost (communication cost and processing cost) for 

processing this query is minimized. The problem is complicated by the fact that query processing not only depends on the 

operations of the query, but also on the parameter values associated with the query. Distributed query processing is an 

important factor in the overall performance of a distributed database system. 

Query optimization is a difficult task in a distributed client/server environment as data location becomes a major factor. In order to optimize queries accurately, sufficient information must be available to determine which data access techniques are most effective (for example, table and column cardinality, organization information, and index availability). Optimization algorithms have an important impact on the performance of distributed query processing. 

In this paper, we describe the distributed query optimization problem in detail. We then present a (ARRQ) technique to process queries with a minimum quantity of intersite data transfer. The technique can be used to process the query where all of the relations referenced by a query are non-fragmented but distributed in different sites. The proposed technique is used to determine which relations are to be partitioned into fragments, and where the fragments are to be sent for processing. The technique is efficient compared to other techniques, as it generally chooses more than one relation to 

remain fragmented which exploits parallelism, while replicating the other relations (excluding the fragmented relations) to the sites of the fragmented relations. Thus the communication costs and local processing costs can be reduced due to the reduced size of the fragmented relations and the response time of queries can be improved. 

Key words: 

Join, Semijoin, Query, FRS, PRS, LR, and Optimization.