The dynamic sub-topology load balancing algorithm for data center networks

Abstract

Many approaches have been suggested for data centers to balance the traffic load nowadays. Both the network topology and routing algorithm can affect the quality of the network communication. Recently, a new network architecture named fat-tree becomes one of the most widely used topologies for data center networks. The routing algorithms, such as global load balancing (GLB) and dynamic load balancing (DLB) algorithms, have been proposed which both use the fat-tree topology for the data centers using SDN architecture, but the GLB and DLB have some limitations of link information storing and path finding between two hosts. In this paper, we present an innovative approach that is called dynamic sub-topology load balancing (DCLB) for fat-tree to schedule network flows in data center networks. The DCLB algorithm overcomes the limitations of DLB algorithms. It will not only use less storage to store link information in the controller, but also choose a path with lowest link cost. Some techniques are taken to achieve the goal of balancing traffic load.