FlowDCN: Flow Scheduling in Software Defined Data Center Networks

Software Defined Networking (SDN) provides intelligence to network with programmability and decoupled layers. If traffic flows are frequently changing in software-defined data center networks then dynamic flow scheduling is required. This paper provides enhancement to the existing dynamic flow scheduling algorithms for Data Center Networks(DCN) by considering both host-based and link-based capacity of the network. The proposed algorithm is implemented on FAT-tree topology with pod 4 (16 hosts as servers) and compared the achieved network performance with the existing dynamic flow scheduling using RYU controller. Compared to Hedera, the proposed algorithm increases 19.33 % and 5% of the average throughput for unsaturated and saturated hot-spot traffic. Also, the proposed algorithm achieved above 75% and 70% of bandwidth utilization for unsaturated and saturated hot-spot traffic which is higher than existing dynamic flow scheduling algorithms.