A traffic splitting method of high-speed network using sub-network flow optimization

The aim of traffic splitting in parallel traffic replay is to distribute high-speed network traffic to multiple replay nodes such that the discrepancies between the generated traffic and the original traffic can be minimized. Currently adopted simulation based methods have a high computational cost and can only guarantee flow level traffic integrity. In this paper, a new method for traffic splitting of high-speed network is proposed based on packet delay time minimization. The problem is converted into a quadratic assignment problem. The computational cost can be dramatically reduced and sub-network level traffic integrity can be achieved by aggregating flows from the same sub-network and assigning traffic in unit of sub-network flows. The method is validated by using actual traffic traces, and a comprehensive comparison between the proposed method and other available traffic splitting methods such as packet based and flow based round-robin approaches is carried out. Empirical studies show that the proposed method can achieve the same replay performance as the flow-based round-robin approach, and achieves 10% performance enhancement over the other approaches with the same split unit.