L. Breslau and S. Shenker. Best-effort versus reservations: a simple comparative analysis. In Proc. ACM SIGCOMM, September 1998.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....an amount of received goods or service, and in the case of bandwidth utility function, the allocated bandwidth. Bandwidth utility functions have been used as a conceptual abstraction of the application s valuation of available bandwidth in analyzing some fundamental Internet design questions [17] [3]. Breslau and Shenker [3] use several types of bandwidth utility functions to investigate the merits of bandwidth reservation for adaptive and rigid applications. The wide adoption of bandwidth utility functions has been impeded, however, by the absent of efficient formulation of ....

....or service, and in the case of bandwidth utility function, the allocated bandwidth. Bandwidth utility functions have been used as a conceptual abstraction of the application s valuation of available bandwidth in analyzing some fundamental Internet design questions [17] 3] Breslau and Shenker [3] use several types of bandwidth utility functions to investigate the merits of bandwidth reservation for adaptive and rigid applications. The wide adoption of bandwidth utility functions has been impeded, however, by the absent of efficient formulation of application specific utility functions. ....

L. Breslau and S. Shenker. Best-effort versus reservations: a simple comparative analysis. In Proc. ACM SIGCOMM, September 1998.

....the evolution of the traffic mix is very hard to forecast. If best effort traffic clearly dominates, then a well provisioned network, possibly enhanced with some simple form of traffic differentiation [3] can probably satisfy the occasional requests for Quality of Service (QoS) guarantees [6]. In other words, appropriate bandwidth is the key to QoS. On the other hand, if the proportion of guaranteed traffic becomes significant, more 0 In Proc. of PfHSN 99, Salem, MA, USA. August 99. J. Sterbenz and J. Touch (Eds. Kluwer Science Publishers. pp 133 150. advanced resource management ....

L. Breslau and S. Shenker. Best-effort versus Reservations: a Simple Comparative Analysis. ACM Comp. Comm. Reviews, 28(4):3--16, Sep 1998.

.... is a continuously differentiable, strictly concave, increasing function in the interval (0; 1) and we assume that U r (x r ) is unbounded as x r 0: Examples of such a function include log x r and Gamma1=x r : An open issue is to incorporate non concave utility functions such as those studied in [3]. Proposition 1: The game admits a unique Nash equilibrium which is also the unique maximum of the following team problem, i.e. one where all users jointly optimize a single performance objective: max fx r g X r w r fi r U r (x r ) Gamma fi X l2L Z X l2j x j 0 (x Gamma C l ....

L. Breslau and S. Shenker. Best-effort versus reservations: a simple comparative analysis. ACM Computer Communications Review, vol. 28, pages 3--16, September 1998.

....TCP window size should take several parameters into account, such as the loss probabilities of in profile and out profile packets, and the sender service profile. Without this, not all of the potential benefits of allowing out profile packets into the network can be achieved. Several studies [2] [3], 13] 17] 19] 27] examine delay and or loss behavior of services in the diffserv architecture using a variety of traffic models. May et al. 17] quantify the expected delay and loss of packets that arrive via a Poisson process, for assured [4] and premium services [20] Naser et al. 19] ....

....[13] consider ways in which assured service can be used to improve TCP throughput, and demonstrate the difficulties in quantifying the impact of assured service on TCP. Yeom et al. 27] perform a similar study, and propose a rather elaborate mechanism to support TCP. Finally, Shenker et al. 2] [3] show that the usefulness of a priority based service depends heavily on the adaptive capabilities of applications. While each piece of work contributes to the understanding of the benefits and limitations of the various services, yet the question that, we believe, is of interest to the ....

L. Braslau, S. Shenker. Best-Effort versus Reservations: A Simple Comparative Analysis, Proceedings of ACM Sigcomm, October 1998, pp. 3-16.

....simplifying assumption in our analysis. Despite these simplifying assumptions, we believe our results show that decision theory can be used in the specification and design of systems. We are not the first to attempt to apply decision theory in computer science. Shenker and his colleagues [BBS98, BS98], for example, have used ideas from decision theory to analyze various network protocols; Microsoft has a Decision Theory and Adaptive Systems group that has successfully used decision theory in a number of applications, including troubleshooting problems with printers and intelligent user ....

L. Breslau and S. Shenker. Best-effort versus reservations: A simple comparative analysis. Submitted for publication, 1998.

....TCP window size should take several parameters into account, such as the loss probabilities of in profile and out profile packets, and the sender service profile. Without this, not all of the potential benefits of allowing out profile packets into the network can be achieved. Several studies [2, 3, 13, 17, 19, 26] examine delay and or loss behavior of services in the diffserv architecture using a variety of traffic models. May et al. 17] quantify the expected delay and loss of packets that arrive via a Poisson process, for assured [4] and premium services [20] Naser et al. 19] quantify the expected ....

....et al. 13] consider ways in which assured service can be used to improve TCP throughput, and demonstrate the difficulties in quantifying the impact of assured service on TCP. Yeom et al. 26] perform a similar study, and propose a rather elaborate mechanism to support TCP. Finally, Shenker et al. [2, 3] show that the usefulness of a priority based service depends heavily on the adaptive capabilities of applications. While each piece of work contributes to the understanding of the benefits and limitations of the various services, yet the question that, we believe, is of interest to the networking ....

L. Braslau, S. Shenker. Best-Effort versus Reservations: A Simple Comparative Analysis, Proceedings of ACM Sigcomm, October 1998, pp. 3-16.

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