A taxonomic framework for autonomous service management in Service-Oriented Architecture

Abstract

Since Service-Oriented Architecture (SOA) reveals the black box nature of services, heterogeneity, service dynamism, and service evolvability, managing services is known to be a challenging problem. Autonomic computing (AC) is a way of designing systems that can manage themselves without direct human intervention. Hence, applying the key disciplines of AC to service management is appealing. A key task of service management is to identify probable causes for symptoms detected and to devise actuation methods that can remedy the causes. In SOA, there are a number of target elements for service remedies, and there can be a number of causes associated with each target element. However, there is not yet a comprehensive taxonomy of causes that is widely accepted. The lack of cause taxonomy results in the limited possibility of remedying the problems in an autonomic way. In this paper, we first present a meta-model, extract all target elements for service fault management, and present a computing model for autonomously managing service faults. Then we define fault taxonomy for each target element and inter-relationships among the elements. Finally, we show prototype implementation using cause taxonomy and conduct experiments with the prototype for validating its applicability and effectiveness.