In this paper we evaluate the effectiveness of Random Early Detection (RED) over traffic types categorized as nonadaptive, fragile and robust, according to their responses to congestion. We point out that RED allows unfair bandwidth sharing when a mixture of the three traffic types shares a link. This unfairness is caused by the fact that at any given time RED imposes the same loss rate on all flows, regardless of their bandwidths. We propose Flow Random Early Drop (FRED), a modified version of RED. FRED uses per-active-flow accounting to impose on each flow a loss rate that depends on the flow's buffer use. We show that FRED provides better protection than RED for adaptive (fragile and robust) flows. In addition, FRED is able to isolate non-adaptive greedy traffic more effectively. Finally, we present a "two-packet-buffer" gateway mechanism to support a large number of flows without incurring additional queueing delays inside the network. These improvements are demonstrated by simulation of TCP and UDP traffic. FRED

Existing approaches for providing guaranteed services require routers to manage per ow states and perform per ow operations [9, 21]. Such a stateful network architecture is less scalable and robust than stateless network architectures like the original IP and the recently proposed Di serv [3]. However, services provided with current stateless solutions, Di serv included, have lower exibility, utilization, and/or assurance level as compared to the services that can be provided with per ow mechanisms. In this paper, we propose techniques that do not require per ow management (either control or data planes) at core routers, but can implement guaranteed services with levels of exibility, utilization, and assurance similar to those that can be provided with per ow mechanisms. In this way we can simultaneously achieve high quality of service, high scalability and robustness. The key technique we use is called Dynamic Packet State (DPS), which provides a lightweight and robust mechanism for routers to coordinate actions and implement distributed algorithms. We present an implementation of the proposed algorithms that has minimum incompatibility with IPv4.

Router mechanisms designed to achieve fair bandwidth allocations, like Fair Queueing, have many desirable properties for congestion control in the Internet. However, such mechanisms usually need to maintain state, manage buffers, and/or perform packet scheduling on a per flow basis, and this complexity may prevent them from being cost-effectively implemented and widely deployed. In this paper, we propose an architecture that significantly reduces this implementation complexity yet still achieves approximately fair bandwidth allocations. We apply this approach to an island of routers -- that is, a contiguous region of the network -- and we distinguish between edge routers and core routers. Edge routers maintain per flow state; they estimate the incoming rate of each flow and insert a label into each packet header based on this estimate. Core routers maintain no per flow state; they use FIFO packet scheduling augmented by a probabilistic dropping algorithm that uses the packet labels an...

Router mechanisms designed to achieve fair bandwidth allocations, like Fair Queueing, have many desirable properties for congestion control in the Internet. However, such mechanisms usually need to maintain state, manage buffers, and/or perform packet scheduling on a per flow basis, and this complexity may prevent them from being cost-effectively implemented and widely deployed. In this paper, we propose an architecture that significantly reduces this implementation complexity yet still achieves approximately fair bandwidth allocations. We apply this approach to an island of routers -- that is, a contiguous region of the network -- and we distinguish between edge routers and core routers. Edge routers maintain per flow state; they estimate the incoming rate of each flow and insert a label into each packet header based on this estimate. Core routers maintain no per flow state; they use FIFO packet scheduling augmented by a probabilistic dropping algorithm that uses the packet labels and an estimate of the aggregate traffic at the router. We call the scheme Core-Stateless Fair Queueing. We present simulations and analysis on the performance of this approach.

The Differentiated services (diffserv) architecture has been proposed as a scalable solution for providing service differentiation among flows without any per-flow buffer management inside the core of the network. It has been advocated that it is feasible to provide service differentiation among a set of flows by choosing an appropriate "markingprofile"for each flow. In this paper, we examine (i) whether it is possible to provide service differentiation among a set of TCP flows by choosing appropriate marking profiles for each flow, (ii) under what circumstances, the marking profiles are able to influencethe service that a TCP flow receives, and, (iii) how to choose a correct profileto achieve a given service level. We derive a simple, and yet accurate, analytical model for determining the achieved rate of a TCP flow when edge-routers use "token bucket"packet marking and core-routers use active queue management for preferential packet dropping. From our study, we observe three important results: (i) the achieved rate is not proportional to the assured rate, (ii) it is not always possible to achieve the assured rate and, (iii) there exist ranges of values of the achieved rate for which token bucket parameters have no influence. We findthat it is not easy to regulate the service level achieved by a TCP flow by solely setting the profileparameters. In addition, we derive conditions that determine when the bucket size influencesthe achieved rate, and rates that can be achieved and those that cannot. Our study provides insight for choosing appropriate token bucket parameters for the achievable rates.

Today’s Internet provides one simple service: best effort datagram delivery. This minimalist service allows the Internet to be stateless, that is, routers do not need to maintain any fine grained information about traffic. As a result of this stateless architecture, the Internet is both highly scalable and robust. However, as the Internet evolves into a global commercial infrastructure that is expected to support a plethora of new applications such as IP telephony, interactive TV, and e-commerce, the existing best effort service will no longer be sufficient. In consequence, there is an urgent need to provide more powerful services such as guaranteed services, differentiated services, and flow protection. Over the past decade, there has been intense research toward achieving this goal. Two classes of solutions have been proposed: those maintaining the stateless property of the original Internet (e.g., Differentiated Services), and those requiring a new stateful architecture (e.g., Integrated Services). While stateful solutions can provide more powerful and flexible services such as per flow bandwidth and delay guarantees, they are less scalable than stateless solutions. In particular, stateful solutions require each router to maintain and manage per flow state on the control path, and to perform per flow classification, scheduling, and buffer management on the data path. Since today’s routers can

In this paper, we study the Assured Service model proposed by Clark and Wroclawski [3, 4]. While existing schemes use service profiles that are defined in terms of absolute bandwidth, it is difficult, if not impossible, to design provisioning algorithms that achieve simultaneously good service quality and high resource utilization for such services with large spatial granularities. We propose an Assured Service model, called LIRA (Location Independent Resource Accounting), in which service profiles are defined in units of resource tokens, rather than absolute bandwidth. The number of resource tokens charged for each in-profile packet is a dynamic function of the path it traverses and the congestion level. Defining service profile in terms of resource tokens allows more dynamic and flexible network control algorithms that can simultaneously achieve high utilization and ensure high probability delivery of in-profile packets. We present an integrated set of algorithms that implement the m...

A novel algorithm for buffer management and packet scheduling is presented for providing loss and delay differentiation for traffic classes at a network router. The algorithm, called JoBS (Joint Buffer Management and Scheduling) , provides delay and loss differentiation independently at each node, without assuming admission control or policing. The novel capabilities of the proposed algorithm are that (1) scheduling and buffer management decisions are performed in a single step, and (2) both relative and (whenever possible) absolute QoS requirements of classes are supported. Numerical simulation examples, including results for a heuristic approximation, are presented to illustrate the effectiveness of the approach and to compare the new algorithm to existing methods for loss and delay differentiation.

A novel algorithm for buffer management and rate allocation is presented for providing loss and delay differentiation for traffic classes at a network router. The algorithm, called JoBS, provides delay and loss differentiation independently at each node, without assuming admission control or policing. Contrary to most existing algorithms, scheduling and buffer management decisions are performed in a single step. Both relative

For the past decade, a lot of Internet research has been devoted to providing different levels of service to applications. Initial proposals for service differentiation provided strong service guarantees, with strict per-flow bounds on delays, loss rates, and throughput, but required high overhead in terms of computational complexity and memory, both of which raise scalability concerns. Recently, the interest has shifted to class-based service architectures with low overhead. However, these newer service architectures only provide weak service guarantees, which do not always address the needs of applications. In this article, we introduce a service architecture that supports strong per-class service guarantees, can be implemented with low computational complexity, and only requires to maintain little state information. A key