This paper presents the “allocated-capacity” framework for providing different levels of best-effort service in times of network congestion. The “allocatedcapacity” framework—extensions to the Internet protocols and algorithms—can allocate bandwidth to different users in a controlled and predictable way during network congestion. The framework supports two complementary ways of controlling the bandwidth allocation: sender-based and receiver-based. In today’s heterogeneous and commercial Internet the framework can serve as a basis for charging for usage and for more efficiently utilizing the network resources. We focus on algorithms for essential components of the framework: a differential dropping algorithm for network routers and a tagging algorithm for profile meters at the edge of the network for bulk-data transfers. We present simulation results to illustrate the effectiveness of the combined algorithms in controlling transmission control protocol (TCP) traffic to achieve certain targeted sending rates.

Schemes based on the tagging of packets have recently been proposed as a low-cost way to augment the single class best effort service model of the current Internet by including some kind of service discrimination. Such schemes have a number of attractive features, however, it is not clear exactly what kind of service they would provide to applications. Yet quantifying such service is very important to understand the benets and drawbacks of the different tagging schemes and of the mechanisms in each scheme (for example how much RIO contributes in the Assured scheme), and to tackle key performance and economic issues (e.g. the difference in tariff between different service classes would presumably depend on the difference in performance between the classes). Our goal in this paper is to obtain a quantitative description of the service provided by tagging schemes. Specically, we describe and solve simple analytic models of two recently proposed schemes, namely the Assured Service scheme ...

This paper presents an overview of the plaNET/ORBIT Gigabit networking system being developed at IBM. It identifies the various network components and their functionality, and describes how the many services provided by the network are supported. The paper focuses on providing a general view on how the different components operate and interact with each other, rather than on giving detailed technical descriptions of their implementations. Instead, pointers to relevant publications are used to allow interested readers to find additional details. 1.0 INTRODUCTION The increasing availability of fiber optic transmission facilities has made Gigabit/second networks feasible. Within the next few years, it is likely that such networks will be widely deployed. They will carry a variety of traffic types such as interactive video, file transfers, multimedia mail, remote visualization, distributed computation, etc. The protocols, switching structures and control algorithms necessary to realize ...

Abstruct- A control architecture for a high-speed packet- [6]) will form the basis for multimedia high-speed networks switched network is described. The architecture was designed that will carry voice, data, and video through a common and implemented as part of the PARIS (subsequently plaNET and set of nodes and links. Real time traffic (e.g., voice and BBNS) networking project at IBM. This high bandwidth network for integrated communication (data, voice, video) is currently video) requires that the route selection function be capable operational as a laboratory prototype. It will also be deployed of guaranteeing for a long period the availability of adequate within the AURORA Testbed that is part of the NSFlDARF'A network resources along the chosen path for a particular traffic gigabit networking program. stream. These streams typically require that a minimal amount The high bandwidth dictates the need for specialized hardware of bandwidth be available to them as long as the stream to support faster packet handling for both point-to-point and multicast connections. A faster and more efficient network control is active. On the other hand, nonreal-time services (such as is also required in order to support the increased number of traditional data services) are much less predictable and must

In this paper we provide the analysis of a ratebased access control scheme in high speed environments based on a buffered leaky bucket algorithm. The analysis is carried out in discrete time which is representative of an ATM environment. For the cell arrivals to the leaky bucket we consider a general discrete Markovian arrival process which models bursty and modulated sources. The key of our analysis is the introduction of the deficit function that allows the reduction of the original problem to a more standard discrete time queueing system with the same arrival process. As an important special case, the detailed analysis of the Binary Markov Source throttled by such rate-based access control schemes is presented. Along with explicit recursions for computation of state probabilities and simple characterization of the asymptotic behavior of the queue build up, some guidelines for the parameter selection of these schemes are provided. Our results indicate that for sources with relatively...

The current Internet assumes the best-effort service model. In this model, the network allocates bandwidth among all the instantaneous users as best it can, and attempts to serve all of them without making any explicit commitments as to bandwidth or delay. Routers keep no state about end host connections, and when congestion occurs, all connections are expected to slow down and achieve a collective sending rate equal to the capacity of the congestion point. As the Internet has transitioned from a research network to a commercial, heterogeneous network, three problems arise. First, an increasing number of real-time applications require some kind of quality of service (QoS) guarantees from the Internet rather than the simple best-effort service. Second, a heterogeneous user base has a variety of different requirements from the network and some users are willing to pay to have their requirements satisfied, and the current Internet service model cannot offer a range of flexible services. ...