Optimization of multi-request single-processor computing

Abstract

Background. The effectiveness of multi-query execution in single-processor computer system databases is considered. One of the methods of increasing the performance of databases of computer systems is the simultaneous execution of several queries that form a multi-query. Methods. Methods of analysis and criterial optimization are used in the paper. Results. The paper analyzes the processing of a conjunctive multi-query (queries are formed by the conjunction of elementary queries, from which a number of elementary queries are repeatedly included in the queries). With the growing importance of operational analytical data processing, the technique of more complex query optimizations has become crucial. In order to be effective, optimizers must adapt to new operators, changes in cost estimation methods, etc. The query execution time is usually used as a criterion for multi-query optimization, while subdividing the time spent working with data in RAM, buffer, and external memory. An optimization plan for conjunctive multi-query execution time when accessing the database based on the ordering of elementary queries is proposed. Two methods of performing a conjunctive multiquery are considered: independently of each other and jointly for unordered and ordered data. It is shown that the execution time of a conjunctive multi-query with joint processing is not always less than with independent processing. The conditions under which the joint processing of a conjunctive multi-query ensures no longer execution time in relation to independent processing are proved. Сonclusions. Implementation of the proposed optimization procedure allows one to increase efficiency of single-processor computer systems that support data bases.