Showing papers presented at "International Workshop on Quality of Service in 2000"

A quality of service architecture that combines resource reservation and application adaptation

Abstract: Reservation and adaptation are two well-known and effective techniques for enhancing the end-to-end performance of network applications. However, both techniques also have limitations, particularly when dealing with high-bandwidth, dynamic flows: fixed-capability reservations tend to be wasteful of resources and hinder graceful degradation in the face of congestion, while adaptive techniques fail when congestion becomes excessive. We propose an approach to quality of service (QoS) that overcomes these difficulties by combining features of reservations and adaptation. In this approach, a combination of online control interfaces for resource management, a sensor permitting online monitoring, and decision procedures embedded in resources enable a rich variety of dynamic feedback interactions between applications and resources. We describe a QoS architecture, GARA, that has been extended to support these mechanisms, and use three examples of application-level adaptive strategies to show how this framework can permit applications to adapt both their resource requests and behavior in response to online sensor information.

Proportional differentiated services, part II: loss rate differentiation and packet dropping

Abstract: The proportional differentiation model was proposed in Dovrolis et al. (1999), as a target for controllable and predictable relative differentiated services. Only the delay differentiation aspect of the model was considered, and focused on packet scheduling mechanisms. In this paper, we extend the proportional differentiation model in the direction of loss rate differentiation. Several previous mechanisms for buffer management and packet dropping, such as complete buffer partitioning, partial buffer sharing, or multi-class RED, are not suitable for relative differentiated services. We propose and evaluate two dropping mechanisms that closely approximate the proportional loss rate differentiation model. The two droppers, PLR(/spl infin/) and PLR(M), differ in the time interval over which the loss rates are measured and proportionally adjusted. This difference results in several trade-offs, in terms of implementation complexity, accuracy, and ability to deal with nonstationary traffic loads. We also re-evaluate the delay differentiation that the waiting time priorities (WTP) scheduler, of Dovrolis et al., can achieve in the presence of finite buffers and packet losses; this study extends their results providing further insight into the behavior of WTP in heavy load conditions. Finally, we examine the coupled effect of delay and loss rate proportional differentiation on the throughput of bulk-transfer TCP connections.

MLDA: a TCP-friendly congestion control framework for heterogeneous multicast environments

Abstract: To avoid overloading the Internet and starving TCP connections, multimedia flows using non-congestion controlled UDP need to be enhanced with congestion control mechanisms. In this paper, we present a general framework for achieving TCP-friendly congestion control called MLDA. Using MLDA, multimedia senders adjust their transmission rate in accordance with the network congestion state. Taking the heterogeneity of the Internet and the end systems into account, MLDA supports layered data transmission where the shape and number of the layers is determined dynamically based on feedback information generated by the receivers. Performance tests of MLDA as well as comparisons to other congestion control schemes suggest that MLDA achieves TCP-friendly congestion control on the one hand and is able to accommodate the needs of heterogeneous receivers on the other.

Of packets and people: a user-centered approach to quality of service

Abstract: Multimedia communication has gained increasing attention, both from the application side and the network provider side. While resource provisioning for QoS support in packet switched networks has led to the design and development of sophisticated QoS architectures, notably ATM, IntServ or DiffServ, research has not exactly been user or application-context centered. In the cause of the evolution of QoS architectures, the integrated service network approach has lost momentum, and with it, the notion of QoS guarantees. Differentiation of QoS classes within the DiffServ framework is based on the definition of various per-hop behaviors. What is currently missing is a technique for specification and mapping of application and user QoS preferences onto evolving service profiles. In addition, adaptation of applications (and users) is becoming increasingly important in the face of dominating weak QoS-assurance paradigms, both in wireline and wireless environments. As a prerequisite, this paper investigates cognitive and perceptive conditioning of users and applications in a situated setting. The contribution of this paper is twofold: first, essential empirical results on user QoS preferences and QoS graduations are presented, and second, methodological foundations are laid for investigating user-centered QoS.

Multi-class latency-bounded Web services

Abstract: Two recent advances have resulted in significant improvements in Web server quality of service. First, both centralized and distributed Web servers can provide isolation among service classes by fairly distributing system resources. Second, session admission control can protect classes from performance degradation due to overload. The goal of this work is to design a general "front-end" algorithm that uses these two building blocks to support a new Web service model, namely, multi-class services which control response latencies to within pre-specified targets. Our key technique is to devise a general service abstraction to adaptively control not only the latency of a particular class, but also to assess the inter-class relationships. In this way, we capture the extent to which classes are isolated or share system resources (as determined by the server architecture and system internals) and hence their effects on each other's QoS. For example, if the server provides class isolation (i.e., a minimum fraction of system resources independent of other classes), yet also allows a class to utilize unused resources from other classes, the algorithm infers and exploits this behavior without an explicit low level model of the server. Thus, as new functionalities are incorporated into Web servers, the approach naturally exploits their properties to efficiently satisfy the classes' performance targets. We validate the scheme with trace driven simulations.

QoS provisioning using a clearing house architecture

Abstract: We have designed a clearing house (CH) architecture that facilitates resource reservations over multiple network domains, and performs local admission control. Two key ideas employed in this design to make the CH scalable to a large user base are hierarchy and aggregation. In our model, we assume the network is composed of various basic routing domains which can be aggregated to form logical domains. This introduces a hierarchical tree of logical domains and a distributed CH architecture is associated with each logical domain to maintain the intra-domain aggregate reservations. The parent CH in the logical tree maintains the inter-domain reservation requests. Call setup time is reduced by performing advanced reservations based on statistical estimates of the call traffic across various links. We explore, with simulations, the efficiency of the CH-architecture in terms of resource utilization, call rejections and reservation setup time.

Efficient QoS partition and routing of unicast and multicast

Abstract: In this paper we study problems related to supporting unicast and multicast connections with quality of service (QoS) requirements. We investigate the problem of optimal routing and resource allocation in the context of performance-dependent costs. In this context each network element can offer several QoS guarantees, each associated with a different cost. This is a natural extension to the commonly used hi-criteria model, where each link is associated with a single delay and a single cost. This framework is simple yet strong enough to model many practical interesting networking problems. An important problems in this framework is finding a good path for a connection that minimizes the cost whilst retaining the end-to-end delay requirement. Once such a path (or a tree in the multicast case) is found, one needs to partition the end-to-end QoS requirements among the links of the path (tree). We consider the case of general integer cost functions (where delays and cost are integers). As the related problem is NP complete, we concentrate on finding efficient /spl epsi/-approximation solutions. We improve on recent previous results, both in terms of generality as well as in terms of complexity of the solution. In particular, we present novel approximation techniques that yield the best known complexity for the unicast QoS routing problem, and the first approximation algorithm for the QoS partition problem on trees, both for the centralized and distributed cases.

A TCP-friendly traffic marker for IP differentiated services

Abstract: The differentiated services architecture allows the provision of "better-than-best-effort" services in the Internet. However the performance of legacy TCP applications over differentiated services is still influenced by bursty packet loss behavior. This paper proposes the use of TCP-friendly building blocks in the Diffserv architecture, and in particular, a TCP-friendly traffic marker to enhance TCP performance over assured service. We present the marker design and resulting performance improvements in terms of improved predictability of service, reduced provisioning requirements, better fairness and fewer TCP timeouts. Though the marker improves assured service performance predictability compared to a simple token bucket marker, the performance is still dependent to some extent on TCP dynamics.

Distributed QoS compilation and runtime instantiation

Abstract: The rapid growth and coexistence of different application domains, such as multimedia and electronic commerce, present a significant challenge to the provision of their quality of service (QoS). To solve this challenge, we need a unified QoS framework, which allows flexibility and reconfigurability. In this paper, we present a reconfigurable component-based QoS framework, called 2K/sup Q/, which solves the challenge by partitioning the end-to-end QoS setup process into distributed QoS compilation and runtime QoS instantiation phases for different types of applications. Entities, services and protocols of this framework, such as application-to-component translator and component-to-resources translators, achieve the distributed QoS compilation and prepare all necessary QoS structures for the end-to-end QoS setup. Other capabilities of this framework, such as a reconfigurable middleware and functional adaptation, achieve the runtime instantiation of the end-to-end QoS setup. We have implemented the first prototype of this framework and the results show a feasible overhead of the runtime service instantiation and reconfiguration for different applications and their QoS requirements.

A measurement-based analysis of expedited forwarding PHB mechanisms

Abstract: The differentiated services (Diffserv) architecture defines a new framework for the support of quality of service (QoS) in IP-based networks. In this paper, we focus on the mechanisms used to implement the expedited forwarding (EF) per-hop behavior (PHB). We analyze these mechanisms with respect to two main QoS parameters: delay and delay variation. Within a Diffserv node we study the impact of buffering and we compare the effectiveness of several different scheduling algorithms for the treatment of EF traffic. Tests are carried out in different network configurations under a range of different parameters and traffic conditions for the analysis of classification, metering, packet marking, policing and scheduling (WFQ and priority queuing). Our study has as its goal to explore quantitatively and in detail the operation of several different router mechanisms used to implement the EF PHB in production-class routers.

Traffic shaping for end-to-end delay guarantees with EDF scheduling

Abstract: The provision of quality of service (QoS) in terms of end-to-end delay guarantees to real-time applications is an important issue in emerging broadband packet networks. Of the various packet scheduling schemes that have been proposed in the literature, earliest deadline first (EDF) scheduling in conjunction with per-hop traffic shaping (jointly referred to as rate-controlled EDF or RC-EDF) has been recognized as an effective means of end-to-end deterministic delay provisioning. An important aspect that has not been addressed satisfactorily in the literature, however, concerns the choice of RC-EDF shaping parameters that realize maximal network utilizations. In this paper, we first establish that except in trivial cases, it is infeasible to identify "optimal" shapers that realize maximal RC-EDF schedulable regions. Ascertaining the optimal flow shaper requires the state of the entire network to be considered, making it computationally impractical. We then propose an heuristic choice of shaper derived from the number of hops traversed by the flow. The resulting shaper is easy to compute, and varies gracefully between the known optimal shapers for limiting values of the hop-length. We show via simulations that for a realistic traffic mix, our choice of shaper allows RC-EDF to outperform the GPS (generalized processor sharing) scheduling discipline as well as RC-EDF disciplines that use shapers chosen independent of the flow hop-length.

Dynamic edge provisioning for core IP networks

Abstract: Effective edge capacity provisioning is an important architectural component of the emerging differentiated service Internet. We propose a set of dynamic provisioning algorithms, which are operational at the edge routers of a differentiated services core network. These edge mechanisms include: (i) ingress dynamic link sharing, which augments class based queueing techniques with bandwidth utility functions so that dynamic link sharing can be used to distribute bandwidth among traffic conditioners located at edge routers; and (ii) egress dynamic capacity dimensioning, which formulates bandwidth dimensioning at egress links to peering/transit networks taking into account measured core network traffic conditions. We demonstrate through analysis and simulation that the proposed edge provisioning architecture is efficient and effective at supporting user link sharing policies while taking into account the dynamics of traffic load measured in the core network.

Tradeoffs in designing networks with end-to-end statistical QoS guarantees

Abstract: Recent research on statistical multiplexing has provided many new insights into the achievable multiplexing gain in QoS networks, however, generally only in terms of the gain experienced at a single switch. Evaluating the statistical multiplexing gain in a general network remains a difficult challenge. In this paper we describe two distinct network designs for statistical end-to-end delay guarantees, referred to as class-level aggregation and path-level aggregation, and compare the achievable statistical multiplexing gain. Each of the designs presents a particular trade-off between the attainable statistical multiplexing gain and the ability to support delay guarantees. The key characteristic of both designs is that they do not require, and instead, intentionally avoid, consideration of the correlation between flows at multiplexing points inside the network. Numerical examples are presented for a comparison of the two designs. The presented class-level aggregation design is shown to yield very high achievable link utilizations while simultaneously achieving the desired statistical guarantees on delay.

Guaranteeing quality of service to packetised voice over the UMTS air interface

Abstract: A key feature of the air interface of the Universal Mobile Telecommunication System (UMTS) standardised by 3GPP is that its functionality can be tailored to the requirements of individual applications, In this paper we study how this flexibility can be exploited to meet the stringent delay and packet loss requirements of a voice-over-IP application (VoIP) We introduce an analytical model of the UMTS air interface that allows us to quantify the impact of the air interface parameters (power budget, interleaving span and channel coding rate) on the parameters that determine the perceived quality of VoIP calls (mouth-to-ear delay and packet loss probability) We use the model to explore the trade-offs that are inherent in the setting of the air interface parameters The numerical results we present illustrate and quantify the trade-offs of bandwidth efficiency versus packet loss, and of packet loss versus packet delay The latter trade-off is of particular importance for the feasibility of VoIP

MPLS traffic engineering using enhanced minimum interference routing: an approach based on lexicographic max-flow

Abstract: The main contribution of this paper is a new algorithm for dynamic routing of bandwidth guaranteed tunnels. The algorithm is a generalization of the "minimum-interference" based algorithm presented in Kodialam and Lakshman (2000). The routing objective is to satisfy as many potential future requests as possible in an environment where requests arrive one-by-one, and there is no knowledge of future arrivals other than their potential ingress-egress points. An application of these algorithms is for the routing of MPLS bandwidth-guaranteed label-switched paths (LSP). The presented algorithm performs better than the previous scheme, and other previously proposed algorithms, on several metrics like the number of rejected demands and successful re-routing of demands upon link failure. Also, the new algorithm is more robust in performance to implementations where most requests are routed with a shortest-path computation, and the computations required for determining "minimum-interference" are performed only infrequently.

Towards a theory of differentiated services

Abstract: Architecting networks capable of providing scalable, efficient, and fair services to users with diverse QoS requirements is a challenging problem. The differentiated services framework has advanced a set of building blocks comprised of per-hop and access point behaviors. We provide a theoretical framework for reasoning about differentiated services networks, constrained to be implementable in IP networks. The control framework incorporates assumptions, albeit weak, about selfish user behavior and service provider behavior. This is necessitated by the essential role they play in influencing end-to end QoS, without which an effective evaluation of DiffServ architectures remains incomplete. We show that there is an intimate relationship between the properties exported by per-hop and edge control, and the "goodness" of the resource allocation and QoS attained in a noncooperative network environment. Our control framework-Scalar QoS Control-generalizes the per-hop and edge control achievable by setting a scalar value in packet headers, e.g., the TOS field of IP. We develop a theory of optimal classifiers and the properties they exhibit which facilitate end to-end QoS via the joint action of aggregate flow control per hop and per-flow control at the edge. We show the stability and efficiency properties of the overall network system when users are allowed to influence the choice of scalar values in the DS field at the edge, and service providers export costs to users commensurate with the QoS received.

Minimum-latency dual-leaky-bucket shapers for packet multiplexers: theory and implementation

Abstract: The need for integrating per-flow dual leaky-bucket regulators with a scheduler that enforces strict bandwidth guarantees is becoming a crucial issue in networks that provide differentiated services. Despite the increasing relevance of the topic, existing solutions either lack accuracy in their shaping action, or compromise the provision of strict bandwidth guarantees to the shaped flows, or suboptimally utilize the output link. In this paper, we present the monolithic shaper-scheduler (MSS), which integrates the shaping and scheduling functions while preserving the bandwidth guarantees of the shaped flows. The MSS handles shaped and unshaped flows simultaneously. It allows us to provision per flow shaping parameters that are independent of the corresponding bandwidth requirements; thus, the available link bandwidth can be fully utilized. The latency and worst case fairness properties of the new scheduler are optimal among GPS-related schedulers. We describe how the MSS can be easily implemented in practice, with minimal overhead with respect to a conventional GPS-related scheduler.

Impact of resource reservation on the distributed multi-path quality of service routing scheme

Abstract: This paper studies the impact of resource reservation on the multipath quality-of-service (QoS) routing schemes that use the global network state to make routing decisions. Incorporating resource reservation into multipath QoS routing algorithms greatly changes the communication characteristics in a network system and affects the performance of the routing algorithms. In this paper, we develop a QoS routing protocol that combines resource reservation with the ticket-based distributed multipath QoS routing scheme, evaluate the new routing protocol through extensive simulation, and study the impact of other network components, such as the network size and the link state update mechanisms, on the performance of multipath QoS routing schemes with resource reservation.

Scalable service differentiation using purely end-to-end mechanisms: features and limitations

Abstract: We investigate schemes for achieving service differentiation via weighted end-to-end congestion control mechanisms within the framework of the additive increase/multiplicative decrease (AIMD) principle, and study their performance as instantiations of the TCP protocol. Our first approach considers a class of weighted AIMD algorithms. This approach does not scale well in practice because it leads to excessive loss for flows with large weights, thereby causing early timeouts and a reduction in throughput. Our second approach considers a class of loss adaptive weighted AIMD algorithms. This approach scales by an order of magnitude compared to the previous approach, but is more susceptible to short-term unfairness and is sensitive to the accuracy of loss estimates. We conclude that adapting the congestion control parameters to the loss characteristics is critical to scalable service differentiation; on the other hand, estimating loss characteristics using purely end-to-end mechanisms is an inherently difficult problem.

A representation of network node QoS control policies using rule-based building blocks

Abstract: Network node functions, such as QoS or the security functions of routers, are becoming increasingly complex, so programs, not only configuration parameters, are required to control network nodes. In a policy-based network, a policy is defined at a policy server as a set of rules that deploy at network nodes where it must be translated into an executable program or parameters. Thus, a policy must be represented by a form in which the syntax and semantics are clearly defined, and which can be mechanically translated into an executable program. This is possible if the policy is written in an appropriate rule-based programming language. This paper describes such a language in which functions required for DiffServ can be specified for the interface between a policy server and network nodes. In this language, a policy rule can be composed using predefined primitive building blocks and control structures.

A scalable share differentiation architecture for elastic and real-time traffic

Abstract: Today's Internet lacks the mechanisms necessary for traffic isolation (to divide resources fairly) and differentiation (to allocate resources to users according to their willingness to pay). In this work, we present a new architecture for relative service differentiation, scalable share differentiation (SSD), which allocates network resources on a user basis and properly provides for both elastic and real-time traffic isolation and meaningful differentiation. SSD does not require the storage of any per-flow or per-user information at the core nodes and is inherently designed to work without admission control, making it easier to deploy and manage. In addition, we present a way to provide more fine-grained flow control to users without changing the concepts of SSD, which permits users to choose different priorities for their packets/flows. This allows the integration of a novel concept for improving the service quality given to user flows: user-based admission control. With user-based admission control, the network is unaware of admission control decisions, and the user himself can perform the tasks of accepting, rejecting and prioritizing flows.

On mapping of QoS between integrated services and differentiated services

Abstract: End-to-end QoS in IP networks views Intserv, acting on a per-flow basis, at the peripheries of large core networks where Diffserv, acting on aggregation of flows, operates. In this context, it is necessary to map QoS, in terms of QoS parameters, between the individual and the aggregate flow levels. This is crucial as only the individual, Intserv flow performance is relevant to the end user, whereas the aggregate, Diffserv view is of interest to the network operator only. A deterministic approach at the packet level reveals the presence of synchronization that may result in biases of loss and delay among the individual flows at a Diffserv router, and at the edge router where packets are tagged according to their conformance. We show that, qualitatively, packet loss at the Diffserv level results in 'bursty loss' at the Intserv level. Quantitatively, packet loss at the Diffserv region is first related to consecutive, Intserv packet loss and then to the local packet loss at the Intserv level. As for delay, synchronization may lead a packet to experience maximum delay at every network element of the end-to-end path.

Fair queuing with in/out bit in core stateless networks

Abstract: Several fair queuing schemes have been proposed for core stateless networks They can achieve fairer bandwidth allocation than RED (random early detection) However, there have not been any studies that consider in/out bit usage This paper proposes a way to extend the schemes that have been proposed for core stateless networks to allow the support of in/out bit usage The extended schemes are suitable for realizing guaranteed rate (GR) service, which provides each user with a minimum guaranteed rate and a fair share of the residual bandwidth We present the performance of one of the extended schemes and compare the scheme to RIO (RED with in/out bit) in terms of fair bandwidth allocation

A policy based approach for QoS-on-demand over the Internet

Abstract: This paper describes a systematic policy-based approach for offering differentiated services over a single provider's network. Emerging work in this area both in standards forums as well as in research publications, is heavily focused on the element management problem, i.e., on the specification of policy rules for managing one or more devices. However, an ISP faces the more challenging task of having to offer and deploy end-to-end services across the Internet. We take a virtual leased line service as exemplar, and describe in detail a process through which a customer's request for connectivity between multiple sites is translated into network level policies, validated, admitted, resolved into role level policies, and activated on network devices. Deliberately, this is a "big picture" paper, where we bring together various functional components (such as bandwidth brokers, directories and policy servers) and algorithms (route computation, consistency checking, admission control) to fashion a network management framework for QoS-on-demand.

Application-level IP measurements for multimedia

Abstract: The benefits of application-level packet multimedia measurements are described by using VoIP as an example. The need for statistical reliability and comparative QoS analyses is illustrated, and the role of QoS measurements in providing end-to-end QoS via per-domain QoS budgets is discussed.

Service guarantees in the Internet: differentiated services approach

Abstract: Differentiated services (DiffServ) facilitate packet classification in the Internet. In this paper, we develop a service model for an assured forwarding (AF) PHB group that has the notion of "service guarantees". In particular, we extend the "olympic service" model proposed for the AF PHB group to include guarantees on packet drop rates. We propose an enhanced random early detection buffer management scheme, referred to here as AMRED-G, to support service guarantees using the AF PHB group. It is based on buffer dimensioning and adaptive adjustment of drop thresholds.

Integration of QoS guarantees into SMIL and its flexible implementation

Abstract: In this paper, we propose a flexible implementation technique for QoS control mechanisms where they are implemented as separate modules using the constraint oriented style of real time LOTOS. For an application, we define and implement QOS-SMIL with: (1) dynamic switching, and (2) explicit inter-media synchronization facilities. Experimental results show that the proposed technique has some advantages with respect to development cost for QoS control mechanisms.