Guaranteeing Performance in Fault Tolerant Architecture Solution using Software Agent’s Coordination

Abstract

Performance is a critical attribute in evaluating the quality and dependability of service-oriented systems dependent on fault-tolerant architectures. Fault-tolerant architectures have been implemented with redundant techniques to ensure fault-tolerant services. However, replica-related overhead burdens fault-tolerant techniques with associated performance degradation in service delivery, and this consequentially discourages service consumers with discredits for service providers. In this paper, a fault-tolerant approach that adopts replication and diversity was employed on agent-oriented coordination toward guaranteeing the performance of the proposed fault-tolerant architecture solution under a large-scale service request load. In addition, the resultant architecture solution was simulated with Apache JMeter for performance evaluation considering the performability in the absence and presence of a fault load. The simulation experiments and results revealed the architecture’s efficiency in fault tolerance via the timely coordination of logical and replica-related activities by software agents. Noteworthily, the continued service availability and performance were guaranteed for the architecture solution with a significant rate of regularity in the absence and presence of a replica-related fault. Therefore, this study’s performance evaluation methods and results could serve as a veritable milestone for building fault-tolerant service systems with appreciable performability and contribute to the service-oriented fields where performance is inevitable.

Keywords

• web services
• fault-tolerant architecture
• fault tolerance
• performance
• software agents
• service-oriented systems