Towards seamless service migration in network re-optimization for optically interconnected datacenters

摘要

Emerging applications require datacenter networks to provide dynamic and reconfigurable services in order to adapt to the dynamicity of cloud traffic, as well as to maintain the optimality of resource utilization. In this paper, we demonstrate novel re-optimization design techniques for realizing agile and seamless service. Our main contribution is a novel integer linear programming (ILP) based approach which can minimize connection disruption, while optimizing resource utilization in a re-optimization process. We propose an ILP-based approach that integrates resource assignment and resource dependency digraph construction. We also propose a method to determine migration order. We compare the proposed ILP-based approach with a heuristic approach in terms of the number of connection disruptions and the migration process. Simulation results demonstrate the effectiveness of our approach.