An engineering process for autonomous fault management in service-oriented systems

Abstract

Service-Oriented Computing reveals features which are not commonly found in conventional computing paradigms; loose coupling, dynamism, blackbox, evolvability, and heterogeneity. These features make diagnosing and healing faults found in deployed services and service-related elements more challenging than managing conventional systems. Hence, service-oriented systems management often results in problems of increased cost/effort, decreased effectiveness, and irresolvable service faults. Applying key disciplines of autonomic computing to services management would effectively resolve these problems and automate the task. This paper presents a process for managing faults in autonomous manner. We define a whole life-cycle process for managing faults in SOA and the essential artifacts of each phase. Then, we present instructions for carry out each phase. Especially, we propose methods for fault diagnosis and reasoning knowledgebase updates by learning mechanism in detail. The process is not limited to providing a theoretical basis for service management, but it can be practically applied with current service-oriented architecture standards. © 2010 IEEE.