Vasco Freitas

Admission Control in Multiservice IP Networks: Architectural Issues and Trends

The trend toward the integration of current and emerging applications and services in the Internet has launched new challenges regarding service deployment and management. Within service management, admission control has been recognized as a convenient mechanism to keep services under controlled load and ensure the required QoS levels, bringing consistency to the services offered. In this context, this article discusses the role of AC in multiservice IP networks, and surveys current and representative AC approaches. We address and compare the architectural principles of these AC approaches and their main features, virtues, and limitations that have an impact on the quality control of network services. We identify important design aspects that contribute to the successful deployment of flexible and scalable AC solutions in multiservice networks

Distributed Admission Control for QoS and SLS Management

This article proposes a distributed admission control (AC) model based on on-line monitoring to manage the quality of Internet services and Service Level Specifications (SLSs) in class-based networks. The AC strategy covers intra- and interdomain operation, without adding significant complexity to the network control plane and involving only edge nodes. While ingress nodes perform implicit or explicit AC resorting to service-oriented rules for SLS and QoS parameters control, egress nodes collect service metrics providing them as inputs for AC. The end-to-end operation is viewed as a cumulative and repetitive process of AC and available service computation. We discuss crucial key points of the model implementation and evaluate its two main components: the monitoring process and the AC criteria. The results show that, using proper AC rules and safety margins, service commitments can be efficiently satisfied, and the simplicity and flexibility of the model can be explored to manage successfully QoS requirements of multiple Internet services.

Self-adaptive distributed management of QoS and SLSs in multiservice networks

Distributed service-oriented traffic control mechanisms, operating with minimum impact on network performance, assume a crucial role as regards controlling services quality and network resources transparent and efficiently. In this way, we describe and specify a lightweight distributed admission control (AC) model which provides an uniform solution for managing QoS and SLSs in multiclass and multidomain environments. Taking advantage of the consensual need of on-line service monitoring and traffic control at the network edges, AC decisions are driven by feedback from systematic edge-to-edge measurements of relevant QoS parameters for each service type and SLS utilization. This allows self-adaptive service and resource management, while abstracting from network core complexity and heterogeneity. In this paper, introducing an expressive notation, we specify the high-level entities for multiservice provisioning in a domain and formalize service-dependent AC equations to assure both intra and interdomain model operation. A proof-of-concept of the AC criteria effectiveness in satisfying each service class commitments while achieving high network utilization is provided through simulation.

A distributed admission control model for CoS networks using QoS and SLS monitoring

Achieving an admission control strategy for CoS networks covering both intra-domain and end-to-end operation is still an open issue. This paper discusses how AC can be carried out without adding significant complexity to the network control plane and proposes a distributed service-oriented AC model for these networks. The model only involves the network edge nodes leaving the network core unchanged. Ingress nodes perform implicit or explicit service-dependent AC based on both QoS and SLSs utilization metrics, obtained through edge-to-edge online monitoring performed at egress nodes. From an end-to-end perspective, the flow request is used both for AC and available service computation. Relevant aspects of the model interrelated areas and implementation key points are also discussed.

A framework for broadcasting and management of URIs

Abstract Several Internet information services paradigms are based upon links to information resources. In order to achieve interoperability and integration of different services, the Internet community is being dedicating some effort to the standardization of uniform resource identifiers (URI). In this paper, a framework for the management of uniform resource identifiers is discussed. Attention is payed to the issues of broadcasting and registration of original and replicated information resources. The Usenet News and the E-mail distribution lists are considered as the supporting mechanisms.

Measuring QoS in class-based IP networks using multipurpose colored probing patterns

Multiclass IP networks open new dimensions and challenges on active monitoring as efficient strategies of in-band probing are required to sense each class performance without causing noticeable side-effects on real traffic. In our study, we provide new insights on how to perform efficiently active monitoring in these networks, suggesting the use of light and multipurpose probing streams able to capture simultaneously the behavior of multiple QoS metrics of each class. Considering one-way-delay, jitter and loss metrics, we explore different spatial-temporal characteristics of probing, focusing on finding patterns adjusted to each class measurement requirements. We demonstrate that commonly used probing streams fail to capture these metrics simultaneously and we propose novel colored probing patterns able to increase multipurpose active monitoring efficiency. As test environment, we consider a diffserv domain where admission control resorts to feedback from edge-to-edge active monitoring to dynamically control hard real-time, soft real-time and elastic traffic classes. Comparing graphically and statistically the probing and passive measurement outcome of each class, the obtained results show that despite being difficult to match the scale and shape of multiple metrics, a single and properly colored probing stream can capture close and simultaneously the behavior of one-way-delay, jitter and loss, for low in-band probing rates.

Tuning Delay Differentiation in IP Networks Using Priority Queueing Models

This article evaluates the use of Priority Queueing models to achieve delay differentiation in IP networks operating under the Class of Services paradigm. Three models are considered: proportional model, additive model and a novel hybrid schema based on the upper time limit model. The characteristics, behaviour and viability of these models are analysed as regards traffic delay differentiation. The impact of each model on traffic aggregation and on individual flows is also evaluated. This study is complemented by the analysis of delay differentiation from an end-to-end perspective. An adaptive differentiation mechanism is also proposed and discussed.

A distributed admission control model for class-based networks using edge-to-edge QoS and SLS monitoring

The advent of class-based networks has brought new needs for network traffic control in order to assure a certain QoS level. Despite the existing proposals, achieving a generic admission control (AC) strategy for traffic entering these networks is still an open issue. This paper provides new insights on how AC shall be accomplished proposing an encompassing AC model for multi-service class-based networks, which covers both intra-domain and end-to-end operation, without requiring changes in the network core and complex AC signaling. For each service type, AC is distributed and based on both on-line edge-to-edge monitoring of relevant QoS parameters and SLSs utilization. Service monitoring, performed at egress nodes, provides adequate metrics to ingress nodes which take implicit or explicit AC decisions based on service-de pendent criteria. Although being oriented to flow AC. the model can easily be applied to SLS AC. SLS auditing and SLS traffic conditioning are tasks also covered.

A framework for the development of tolerant real-time applications

Abstract This work presents a framework architecture for the development of distributed real-time applications to be integrated into WWW clients. It assumes a WWW environment over networks providing a best-effort delivery service like the internets based on the IP protocol. The framework is that of an application programming interface (API) providing the program developer with the services needed by tolerant real-time applications. Once developed, an application is bundled together with the API to form a WWW plug-in which can subsequently be called from a WWW client interface or browser. The application is then perceived as being integrated into the WWW environment. The design aims to provide real-time applications with a transport service layer assuring near end-to-end isochronism despite the weak guaranties of the underlying network service. The implementation of the mechanisms that allow multistream real-time communications to adapt to the operational conditions of these networks are discussed. In this work, the RTP and RTCP protocols were also implemented as part of the API. Experience with this framework reports the development of a prototype real-time application for multimedia group communication and the analysis of the behaviour of RTP sessions in a real operational situation. The analysis uses protocol state data logged during their operation.

A new path probing strategy for inter-domain multicast routing

Many already in use applications require the provision of QoS services from the underlying network infra-structure. This is particularly true for multicast, since it involves many participants at very sparse locations usually aiming to receive or send multimedia real-time streams. One way to provide QoS is through routing, since QoS aware multicast routing protocols can find feasible multicast trees. In this paper an inter-domain QoS multicast routing protocol is presented, specifically designed for the hierarchical inter-domain scenario, where requirements like intra-domain independency and policy awareness should be met. Emphasis is given to the path probing mechanism used to connect new members to the multicast tree, stressing how it differs from others. Simulation results show that despite using a less aggressive and simplified probing mechanism-more suitable for inter-domain scenarios-the proposed routing strategy can build multicast trees with metrics similar to those build by more aggressive techniques, with considerable less effort.