An integrated support system for design of distributed databases
Uday R. Kulkarni · Medical Entomology and Zoology · 1990
A methodology for designing distributed data bases is presented. The design process, from the data base administrator's point of view, consists of forming and placing data fragments across the geographically dispersed locations. It is recognized that the number of unique ways in which data can be partitioned and allocated across sites is innumerable. Also, related issued such as query processing and concurrency overheads have a bearing on the evaluation of a candidate partitioning and allocation schema and hence have to be made part of the design process. Since the complexity of the integrated problem makes the problem ill-structured, a tool in the form of a Decision Support System is needed to solve the problem. The methodology presented for the design process incorporates such a tool. The design process is viewed as a decision-making process on the part of the designer who is supported by an interactive design support system to find an acceptable partitioning and allocation choice. The support system uses the semantic knowledge about the data base and the human judgement of the designer to generate a restricted number of meaningful candidate alternatives which are evaluated by a query processing algorithm. A few, more promising, alternatives are evaluated in detail using a simulation model. A new query processing algorithm which maximizes parallelism in processing is presented. A simulation model incorporating a 2-phase locking concurrency control mechanism is developed. The design system is implemented and tested using a real-life case study.