Network Fabric Simulation and Validation for Data Center Routing Convergence Under Large-Scale Failure Scenarios
B. Wang, Zimeng Wang, Wenyu Zhao, Yang Liu
Abstract
Modern data centers serve as the backbone of cloud computing infrastructure, supporting millions of concurrent users and mission-critical applications that demand exceptional reliability and performance. The network fabric within these facilities represents a complex interconnection of switches, routers, and links that must maintain seamless connectivity even during catastrophic failure events. Network fabric simulation (NFS) has emerged as an essential methodology for evaluating routing convergence behavior under various failure scenarios before deployment in production environments. Software-defined networking (SDN) technologies have revolutionized how data center operators manage and orchestrate network resources, enabling programmable control planes that can rapidly respond to topology changes. This review examines the current state of simulation frameworks, validation methodologies, and convergence analysis techniques employed in modern data center networks. We analyze how discrete event simulation (DES) and hybrid simulation approaches enable realistic modeling of large-scale failures affecting multiple network components simultaneously. The review covers critical aspects including fabric topology designs, routing protocol implementations, failure detection mechanisms, and performance metrics used to assess convergence speed and stability. We synthesize findings from recent research on Byzantine fault tolerance (BFT) mechanisms, multipath routing strategies, and traffic engineering solutions that enhance network resilience. Particular emphasis is placed on validation techniques that ensure simulation results accurately reflect real-world network behavior under stress conditions. The analysis reveals that while significant progress has been made in simulation accuracy and scalability, challenges remain in modeling complex interactions between control plane and data plane during cascading failures