Ilias Andrikopoulos

A management and control architecture for providing IP differentiated services in MPLS-based networks

As the Internet evolves toward the global multiservice network of the future, a key consideration is support for services with guaranteed quality of service. The proposed differentiated services framework is seen as the key technology to achieve this. DiffServ currently concentrates on control/data plane mechanisms to support QoS, but also recognizes the need for management plane aspects through the bandwidth broker. In this article we propose a model and architectural framework for supporting DiffServ-based end-to-end QoS in the Internet, assuming underlying MPLS-based explicit routed paths. The proposed integrated management and control architecture will allow providers to offer both quantitative and qualitative services while optimizing the use of underlying network resources.

IP ROUTING ISSUES IN SATELLITE CONSTELLATION NETWORKS

SUMMARY The growth in use of Internet-based applications in recent years has led to telecommunication networks transporting an increasingly large amount of Internet Protocol (IP)-based traffic. Proposed broadband satellite constellation networks, currently under development, will be required to transport IP traffic. A case can be made for implementing IP routing directly within the constellation network, in order to transport IP traffic well and to provide good support for IP multicast and for emerging IP-based Quality of Service (QoS) guarantees. This paper examines strategies for implementing and operating IP routing effectively within satellite constellation networks, given known constraints on the constellation resulting from satellite mobility, global visibility, routing and addressing.

An architectural framework for providing QoS in IP differentiated services networks

As the Internet evolves, a key consideration is support for services with guaranteed quality of service (QoS). The proposed differentiated services (DiffServ) framework, which supports aggregate traffic classes, is seen as the key technology to achieve this. DiffServ currently concentrates on control/data plane mechanisms to support QoS but also recognises the need for management plane aspects through the bandwidth broker (BB). In this paper we propose a model and architectural framework for supporting end-to-end QoS in the Internet through a combination of both management and control/data plane aspects. Within the network we consider control mechanisms for traffic engineering (TE) based both on explicitly routed paths and on pure node-by-node layer 3 routing. Management aspects include customer interfacing for service level specification (SLS) negotiation, network dimensioning, traffic forecasting and dynamic resource and routing management. All these are policy-driven in order to allow for the specification of high-level management directives. Many of the functional blocks of our architectural model are also features of BBs, the main difference being that a BB is seen as driven purely by customer requests whereas, in our approach, TE functions are continually aiming at optimising the network configuration and its performance. As such, we substantiate the notion of the BB and propose an integrated management and control architecture that will allow providers to offer both qualitative and quantitative QoS-based services while optimising the use of underlying network resources.

A fair traffic conditioner for the assured service in a differentiated services Internet

A research issue under investigation in the context of differentiated services (DiffServ) is the fair distribution of bandwidth between aggregates sharing the same assured forwarding (AF) class. Multiplexing both responsive and unresponsive flows, e.g. TCP and UDP respectively, leads to unfair sharing of the available bandwidth in over-provisioned networks. To date, much effort has concentrated on experiments using different methods for mapping TCP and UDP flows of the same AF class to the three possible drop precedences of the AF specification. Although this approach may protect responsive from unresponsive flows, it has not been shown to provide adequate fairness. We present a traffic conditioner able to provide fairness between responsive and unresponsive flows originating from the same customer network, using a fair two-rate three-color marker. Its capability for fairness is based on the use of the FRED fair active buffer algorithm to control the token allocation of the token buckets residing in the traffic conditioner. We also show that by employing fair multiple RED (FMRED) at the DiffServ domain ingress node, the overall fairness of the customer network aggregates is improved when compared to the case where the vanilla MRED algorithm is used.

Supporting differentiated services in MPLS networks

Multi-protocol label switching is a relatively new technology based on the association of labels with routes and the use of labels to forward packets. In other words MPLS integrates the label-swapping paradigm with network-layer routing. Differentiated Services define a model for implementing scalable differentiation of QoS in the Internet. Packets are classified and marked, policed and shaped at the edge of the network in order to receive a particular per-hop forwarding behaviour on nodes along their path. The per-flow state does not need to be maintained in the interior network nodes, thus leading to increased scalability. This obviates the use of complex signalling protocols like RSVP. The inherent characteristics of MPLS make it a very good candidate for providing differentiated services. In this paper we describe various approaches which can be used to support differentiated services in MPLS environments.

Providing rate guarantees for internet application traffic across ATM networks

The TCP/IP protocol suite is the standard requirement for all applications that need to communicate over the Internet. As TCP/IP applications are unable to specify the QoS parameters needed for most asynchronous transfer mode (ATM) services, they tend to use the unspecified bit rate (UBR) service category when running across ATM networks. The UBR service utilizes any bandwidth that is left unused by the rest of the ATM services. This has led the ATM forums traffic management group to define a new service category called guaranteed frame rate (GFR). GFR is intended to provide minimum cell rate guarantees and fair access to excess bandwidth left over from higher-priority services. This article first presents a tutorial overview of GFR and then presents a survey of the research work that has been carried out toward the design and implementation of associated ATM switch mechanisms.

Experiments and enhancements for IP and ATM integration: the IthACI project

IthACI has been a European project in the ACTS framework concentrating on fast layer 2 forwarding methods for IP traffic based on labeled flow mechanisms. The approach is also known as IP switching and is considered promising for enhancing IP performance. Several flavors of IP switching have been proposed by various vendors (e.g., IP switching by Ipsilon, Tag Switching by Cisco, ARIS by IBM, IPSOFACTO by the NEC), all of them different and not interoperable. IP switching has been adopted by the IETF under the umbrella of multiprotocol label switching (MPLS). Although MPLS has made remarkable progress, a number of issues remain largely open for further investigation. The scope of the IthACI project was to address such issues and propose solutions. The issues addressed were multicast, QoS, resource management, and mobility support in a multicast environment. IthACI conducted both theoretical and experimental work. Three network islands, each based on a different flavor of IP switching, were set-up and the interoperability of these different IP switching/MPLS flavors were investigated and demonstrated.

A study of TCP/IP throughout performance in heterogeneous ATM local area networks

This paper investigates the maximum throughput performance of TCP/IP over ATM-based local area networks. The testbeds for the measurements use second- generation ATM switches to interconnect various personal computers and workstations in a heterogeneous environment. The effects of parameters such as socket buffer size and message size on TCP/IP performance over ATM adaptation layer 5 are studied. The results are compared with the calculated theoretically available maximum bandwidth offered by ATM, and an estimation of the limits caused by the processing power of the computers involved. Explanations for the observed behaviour are given.