Abstract The rapid growth of cloud computing and Software-Defined Networking (SDN) has necessitated efficient resource management to handle dynamic traffic demands in large- scale datacenters. This paper proposes a distributed framework for joint load balancing and Virtual Network Function (VNF) scaling to mitigate overload and underload condi- tions in SDN-based cloud environments. By leveraging the Alternating Direction Method of Multipliers (ADMM) and heuristic optimization techniques, our approach minimizes deployment and forwarding costs while ensuring efficient resource utilization. We formu- late the problem using Mixed Integer Linear Programming (MILP) and relax it into linear programming (LP) models to reduce computational complexity. Performance evaluations demonstrate that our method achieves faster convergence and lower resource overhead compared to centralized approaches, validated through simulations on k-fat-tree topolo- gies. Our findings highlight the potential of distributed optimization for scalable and robust network management in modern datacenters.
Zhang et al. (Thu,) studied this question.