Design of intelligent traffic load balancing (ITLB) mechanism for software defined data center (SDDC)

The reinforcement learning strategy is adopted for software defined data centers, since it involves software agents to accomplish decision making through continuous learning of environment during drastic traffic. In this research work, we are proposing an intelligent routing algorithm by considering controller capacity and link capacity as the routing metrics. In addition, our approach takes care of migrating switches whenever a controller is over utilized. Distinct reinforcement learning agents are used to handle path computation and switch migration phase individually. Our key objective is to provide intelligent traffic load balancing approach in software defined data center using two distinct software agents. We observe that our proposed approach has achieved nearly 1.2 to 2.1 milliseconds of less latency, 2.5 milliseconds of less response time and low packet loss percentage around 1 to 2.5% and hence overall network throughput improvement is about attained 30 to 50% than other existing approaches through simulation.