A Genetic Algorithm-Based QoS Analysis Tool for Reconfigurable Service-Oriented Systems

Abstract

Reconfigurability is an important requirement in many application systems. Many approaches have been proposed to achieve static/dynamic reconfigurability. Service-oriented architecture offers a certain degree of reconfigurability due to its support in dynamic composition. When system requirements change, new composition of services can be determined to satisfy the new requirements. However, analysis, especially QoS based analysis, is generally required to make appropriate service selections and service configurations. In this chapter, we discuss the development of QoS-based composition analysis techniques and propose a QoS specification model. The specification model facilitates QoS-based specification of the properties of the Web services and the requirements of the application systems. The composition analysis techniques can be used to analyze QoS tradeoffs and determine the best selections and configurations of the Web services. We develop a composition analysis framework and use the genetic algorithm in the framework for composition decision making. The framework currently supports SOA performance analysis. The details of the genetic algorithm for the framework and the performance analysis techniques are discussed in this chapter.