Skip to Main Content
Large-scale computing applications utilizing many computational nodes via networks have become executable because of high-speed network and computing technologies. Many such computing applications repeat calculations and synchronous data exchange among computational nodes alternately, so that much data is transferred among nodes at almost the same time, and network congestion occurs as a result. When we evaluate the performance of networks which accommodate bursty traffic in high-speed networks, conventional simulation techniques cannot observe the behavior and fluctuation of traffic. In this paper, we introduce a new hybrid-system model and use the improved model to observe the behavior of bursty traffic in high-speed networks. Through a computer simulation, we show that the proposed model can observe the behavior of the bursty traffic in high-speed networks in comparison with a conventional hybrid-system model which cannot.
Date of Conference: 23-26 Aug. 2011