An empirical assessment of autonomicity for autonomic query optimizers using fuzzy-AHP technique

摘要

Quality assurance and evaluation has always been a key cause of concern for software developers. This problem has been further aggravated by the complete dependence of business enterprises, financial institutions and stock markets on computer hardware and software. It is therefore needed to propose and develop such software evaluation and quality assurance techniques that can fit into the business model's domain and satisfy the customer needs and aspirations. Autonomic computation is an artificial intelligent based approach used to design and develop software systems which can fit into business model and also satisfy customer needs. These systems are built with self-managed policy system. To guarantee their customers a Total Quality Assurance on the business applications being developed, the paper presents some key aspects of domain-specific software and its quality estimation parameter. In this paper, the authors have analyzed the various aspects of quality metrics of autonomic computation suggested by enhanced ISO 9126 quality model. A universally acceptable approach to assure quality for autonomic computing system would be to measure the Autonomicity of a system to determine whether it is autonomic or not. If it is autonomic then "to what extent'' is the next question? Autonomicity is an excellent indicator to assure quality of the autonomic software. The approach taken to measure the subjective attribute of Autonomicity is fuzzy theory with Analytic Hierarchy Process (AHP) integrated in it. Human assessment is qualitative and fuzzy technique is best candidate to quantify their opinions. For empirical analysis, three different query optimizers are examined to measure autonomicity. The result of the empirical analysis will be validated using the already proposed results of the research studies. The present study will provide a base for further research in terms of development of applications with autonomic features. It will also help in proposing new metrics for quality characteristics to estimate the overall quality of such application. (C) 2020 Elsevier B.V. All rights reserved.