Measurement and performance evaluation of NFS traffic in ATM network

This article has three parts. At first, we explore traffic characterization of an ATM local area network and find the traffic presents both long and s hort range dependence in nature. Secondly, one long range dependent traffic model (fast fractional Gaussian noise) is constructed to explore influence of long-range dependence on the queuing performance of an infinite buffer- size queue. At last, we combine importance sampling scheme and long range d ependent traffic models to simulate a finite buffer-size queue. The results show that both LRD and SRD play an important role in the queuing systems. Especially when LRD is strong, the more buffer solution is almost useless t o improve the queuing performance. Multiplexing and enough capacity are bet ter solutions for queuing performance. (C) 1999 Elsevier Science B.V. All r ights reserved.