K. Park, G. Kim, and M.Crovella. On the Effect of Traffic SelfSimilarty on Network Performance. In Proceedings of the SPIE International Conference on Performance and Control of Network Sys- tems, 1997.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....and thus not amenable to the statistical multiplexing techniques currently found on the Internet. Additional studies claim that the heavytailed distributions of file size, packet interarrival, and transfer duration fundamentally contribute to the self similar nature of aggregate network traffic [10, 11, 16]. While these heavy tailed distributions may contribute to self similarity, we will illustrate that TCP itself is the primary source of self similarity and that this behavior may have dire consequences in computational grids as WAN speeds increase into the gigabit per second (Gb s) range. In ....

....streams. If the c.o.v. is small, the amount of traffic com ing into the gateway in each RTT will concentrate mostly around the mean, and therefore will yield better performance via statistical multiplexing. For purposes of comparison, we also use the Hurst parameter, H, from self similar modeling [7, 10, 11, 12, 16]. While H may be better at determining long term buffer requirements (i.e. large time granularities) it does not provide insight into how well statistical multiplexing performs, i.e. at small time granularities such as a round trip time (RTT) 3 Experimental Study To understand the dynamics ....

K. Park, G. Kim, and M.Crovella. On the Effect of Traffic SelfSimilarty on Network Performance. In Proceedings of the SPIE International Conference on Performance and Control of Network Sys- tems, 1997.

....respect to its mean. This normalizes the spread of a distribution and allows for the comparison of spreads (or coefficientofvarations) over a varying number of communication streams. Rather than use the Hurst parameter from self similar modeling as is done in many studies of network traffic [4, 5,6,15,16], we use the coefficientofvariation (c.o.v. because it better reflects the predictability of the incoming traffic, and consequently, the effectiveness of statistical multiplexing over the Internet. If the c.o.v. is small, the amount of traffic coming into the gatewayineachRTT will concentrate ....

....the mean, and therefore, will yield better performance via statistical multiplexing. 3 Simulation Study The goal of this simulation study is to understand the dynamics of how TCP modulates applicationgenerated traffic. While this issue has been largely ignored in the self similar literature [4, 5, 6, 15,16], weintend to isolate and understand the TCP modulation so that wemay be better able to schedule network resources. Understanding how TCP modulates traffic can have a profound impact on the coefficientofvariation (c.o.v. and hence, throughput and packet loss percentage of network traffic. This, ....

K. Park, G. Kim, and M. Crovella, "On the Effect of Traffic Self-Similarty on Network Performance, " Proceedings of the SPIE International Conference on Performance and Control of Network Systems, 1997.

....and thus not amenable to the statistical multiplexing techniques currently found on the Internet. Additional studies claim that the heavytailed distributions of file size, packet interarrival, and transfer duration fundamentally contribute to the self similar nature of aggregate network traffic [10, 11, 16]. While these heavy tailed distributions may contribute to self similarity, we will illustrate that TCP itself is the primary source of self similarity and that this behavior may have dire consequences in computational grids as WAN speeds increase into the gigabit per second (Gb s) range. In ....

....streams. If the c.o.v. is small, the amount of traffic coming into the gateway in each RTT will concentrate mostly around the mean, and therefore will yield better performance via statistical multiplexing. For purposes of comparison, we also use the Hurst parameter, H , from self similar modeling [7, 10, 11, 12, 16]. While H may be better at determining long term buffer requirements (i.e. large time granularities) it does not provide insight into how well statistical multiplexing performs, i.e. at small time granularities such as a round trip time (RTT) 3 Experimental Study To understand the dynamics ....

K. Park, G. Kim, and M.Crovella. On the Effect of Traffic SelfSimilarty on Network Performance. In Proceedings of the SPIE International Conference on Performance and Control of Network Systems, 1997.

....respect to its mean. This normalizes the spread of a distribution and allows for the comparison of spreads (or coefficientofvarations) over a varying number of communication streams. Rather than use the Hurst parameter from self similar modeling as is done in many studies of network traffic [4, 5,6,15,16], we use the coefficientofvariation (c.o.v. because it better reflects the predictability of the incoming traffic, and consequently, the effectiveness of statistical multiplexing over the Internet. If the c.o.v. is small, the amount of traffic coming into the gatewayineachRTT will concentrate ....

....the mean, and therefore, will yield better performance via statistical multiplexing. 3 Simulation Study The goal of this simulation study is to understand the dynamics of how TCP modulates applicationgenerated traffic. While this issue has been largely ignored in the self similar literature [4, 5, 6, 15,16], weintend to isolate and understand the TCP modulation so that wemay be better able to schedule network resources. Understanding how TCP modulates traffic can have a profound impact on the coefficientofvariation (c.o.v. and hence, throughput and packet loss percentage of network traffic. This, ....

K. Park, G. Kim, and M. Crovella, "On the Effect of Traffic Self-Similarty on Network Performance, " Proceedings of the SPIE International Conference on Performance and Control of Network Systems, 1997.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC