정보기술과데이타베이스저널 (The Journal of Information Technology and Database)
- 제7권2호
- /
- Pages.47-59
- /
- 2000
- /
- 1226-3559(pISSN)
데이터베이스 색인선택 문제에 대한 Davis-Putnam 기반 최적화 알고리즘의 성능 분석
Analyzing the Performance of a Davis-Putnam based Optimization Algorithm for the Index Selection Problem of Database Systems
초록
In this paper, we analyze the applicability of a general optimization algorithm to a database optimization problem. The index selection problem Is the problem to choose a set of indexes for a database in a way that the cost to process queries in the given workload is minimized subject to a given storage space restriction for storing indexes. The problem is well known in database research fields, and many optimization and/or heuristic algorithms have been proposed. Our work differs from previous research in that we formalize the problem in the form of non-linear Integer Programming model, and investigate the feasibility and applicability of a general purpose optimization algorithm, called OPBDP, through experiments. We implemented algorithms to generate workload data sets and problem instances for the experiment. The OPBDP algorithm, which is a non-linear 0-1 Integer Programming problem solver based on Davis-Putnam method, worked generally well for our problem formulation. The experiment result showed various performance characteristics depending on the types of decision variables, variable navigation methods and ocher algorithm parameters, and indicates the need of further study on the exploitation of the general purpose optimization techniques for the optimization problems in database area.
키워드