Model-Based Performance Evaluation of Dispatcher-Based Server Clusters
Junjun Zheng, Hiroyuki Okamura, Tadashi Dohi · 2024
Web server clusters have become the backbone of Internet infrastructure, supporting a vast array of critical applications ranging from e-commerce to cloud computing. Their performance is crucial, as it directly impacts user experience and service quality. Thus, a quantitative performance evaluation of these systems presents a significant challenge. This paper introduces a promising approach to this challenge by using a multi-state system (MSS)-based hierarchical modeling technique combined with a multi-terminal multi-valued decision diagram (MtMDD) representation. Particularly, this paper evaluates the performance of a dispatcher-based server cluster under random load balancing strategies, providing insights into their effectiveness and efficiency.