SDN Based Dynamic Resource Scheduling for Large Scale Data Centers

With the emergence of cloud computing, data centers become the core component of the underlying infrastructure. Operations of the data centers rely on resource availability, bandwidth, and efficient resource scheduling and allocation. SDN based traffic flow management approach can be used for efficient resource scheduling in the data centers by analyzing flow properties. In the past, queuing systems are being used for the assessment of the client/server model for distributed processing. In this paper, a queuing model-based algorithm is proposed for the dynamic resource scheduling for the optimization of resource requirement in data centers. The proposed algorithm takes advantage of SDN traffic flow management technique and finds the optimal number of servers for a data center to reduce the operational cost of the system. The proposed model reduces the required number of servers in the system to achieve a service throughput and is able to cater to various type of requests. In this work a single M|M|c system model is divide into multiple M|M|c sub-systems to optimize the required number of servers. And incoming traffic is divided into different service classes. This model is tested for SDN and the simulation results validate the efficacy of the proposed model.