Traffic load balancing in data center networks: A comprehensive survey

Abstract

The rapid growth in scale and complexity of data centers has established effective load balancing as a critical requirement for optimizing network performance, resource utilization, and service quality. This survey provides a comprehensive examination of load balancing strategies in data center networks, addressing challenges, established techniques, and emerging trends. The study begins with fundamental concepts of data center networks, encompassing network topologies, load balancing definitions, granularity considerations, and inherent challenges. A systematic analysis of load balancing schemes follows, categorized into centralized and distributed approaches. Within centralized schemes, the analysis encompasses both machine learning-based approaches and traditional methodologies, while distributed schemes are examined through single-granularity and mixed-granularity implementations. The survey extends beyond previous reviews by offering comparative analyses of load balancing strategies, a systematic evaluation of performance metrics, and an assessment of their applicability across diverse data center scenarios. The discussion further explores future development trends driven by technological innovations across multiple dimensions, including hardware, software, architecture, and algorithms. This comprehensive technical review serves as a foundational resource for researchers and practitioners, facilitating advancements in efficient, scalable, and user-centric load balancing technologies for data center networks.