SD-WAN ARCHITECTURAL MODELS: A LITERATURE REVIEW OF CENTRALIZED, DISTRIBUTED, AND HYBRID CONTROL PLANE DESIGNS
Keywords:
Software Defined Wide Area Networking, Artificial Intelligence (AI), Security ThreadsAbstract
Software-Defined Wide Area Networking (SD-WAN) has emerged as a transformative solution for modern network management, offering agility, cost-efficiency, and enhanced security over traditional WAN architectures. This paper presents a systematic literature review of SD-WAN control plane designs, categorizing them into centralized, distributed, and hybrid models. The centralized approach, governed by a single controller, simplifies policy enforcement but introduces scalability bottlenecks and single-point-of-failure risks. Distributed architectures mitigate these issues by decentralizing control, improving resilience at the cost of synchronization complexity. Hybrid models strike a balance, combining global policy coordination with localized decision-making. Through a rigorous methodology—encompassing keyword-based searches, inclusion/exclusion criteria, and thematic synthesis—we analyze 1) the trade-offs between these architectures, 2) their performance under scalability and security threats (e.g., DDoS attacks, controller compromises), and 3) emerging mitigation strategies (e.g., clustering, zero-trust frameworks). Key findings reveal that while centralized designs dominate enterprise deployments, hybrid models are gaining traction for multi-cloud environments. The review also identifies gaps in standardized security protocols and AI-driven dynamic control plane adaptation, suggesting future research directions. This work provides a foundational reference for network architects and researchers evaluating SD-WAN design paradigms