Design and Evaluation of a Fault-Tolerant Spine-Leaf Data Center Network Using Cisco Packet Tracer

Abstract

This study presents the design and performance evaluation of a fault-tolerant spine-leaf data center network for higher education institutions, implemented and tested using Cisco Packet Tracer. Targeting Abdu Gusau Polytechnic Talata Mafara as a case study, the architecture leverages redundant spine switches (Cisco 3560) and dual-homed leaf switches (Cisco 2960) with Hot Standby Router Protocol (HSRP) and Rapid Spanning Tree Protocol (RSTP) to ensure high availability. Through controlled simulations of link/router failures, the network demonstrated sub-second failover times (HSRP: 400ms, RSTP: 1.2s) and maintained operational stability with only an 18% latency increase during failures outperforming traditional three-tier designs by 17%. Key metrics including throughput (940Mbps baseline, 4.8% degradation during failures) and bandwidth utilization (40% normal, 55% failure conditions) validate the design's resilience. The study contributes: (1) empirical evidence of spine-leaf viability for resource-constrained institutions, (2) a benchmarked configuration achieving 99.98% availability with off-the-shelf hardware, and (3) identified optimization opportunities through TRILL/SPB protocols to address the observed 5% throughput drop during re-convergence. These findings provide network architects with a proven template for academic data centers, balancing cost-efficiency with enterprise-grade fault tolerance.