DTE:Dynamic Traffic Engineering in Software Defined Data Center Networks

Abstract

Nowadays, data center networks confront a huge amount of data that can cause both network congestion and packet loss; therefore, traffic engineering methods can help to balance the load through the network. In recent years, quite a bit of traffic engineering methods have been proposed in order to reduce network utilization, especially in cloud data center networks. Reducing network utilization; preventing network congestion, which leads to guaranteeing QoS; and optimal using of the existing route are considered as major challenges through all these works. Prevalent traffic engineering algorithms such as ECMP do not have any focus on the current network circumstance, nor do they provide a solution for mice flows. In this work, we propose a novel dynamic traffic engineering method in software-defined data center networks which considers current network circumstance, uses network resources in an optimal manner, and guarantees QoS. The algorithm uses OpenFlow protocol to detect new flow, gather network information in short intervals, and choose the best route for the flow based on network loads. The proposed algorithm selects the best path through the network for each flow based on their existing flows' type in order to not only improve the QoS but also achieve more customer satisfaction. The evaluation results demonstrate that the DTE algorithm reduces high-priority flows jitter, increase network utilization, and balance loads through the network and path to reach hosts better in comparison with existing traffic engineering methods.