A performance optimization technique for an object-oriented functional data model

摘要

This paper describes a technique to implement an object-oriented functional data model that optimizes performance for a given set of applications (workload). The technique involves grouping functions that are often referenced together or that are often composed. These groups are referred to as access trees which may be considered as abstract storage structures. The focus of this paper is the design of a set of access trees (called the cover) for a given conceptual schema so that the cost of processing the workload is as low as possible and the space of the cover is within some user-specified limit. It is found that using this implementation scheme, a significant reduction in access time is obtainable at the cost of a tolerable space increment. The paper describes the conceptual data model, the structure and use of access trees, an implementation of access trees, costs associated with operations on the access trees, and a methodology to design the optimal set of access trees for a conceptual schema with respect to a given workload. An experimental analysis of the performance of the technique is presented.