Affordable Access

Engineering a Quality of Service-Enabled Network: a Case Study

Authors
Publisher
IEEE
Publication Date
Disciplines
  • Engineering

Abstract

paper.doc ENGINEERING A QUALITY OF SERVICE-ENABLED NETWORK: A CASE STUDY Fulvio Risso Dipartimento di Automatica e Informatica, Politecnico di Torino, Corso Duca degli Abruzzi, 24, 10129 Torino - Italy ABSTRACT This paper presents a case study concerning a network operator that wants to provide services with a certain degree of quality to its customers. Users are presented with a set of services with different Quality of Service capabilities and the network is engineered in order to satisfy the requests and to maximize provider’s revenues. This paper presents the service model adopted and the technical choices that allowed this network to work, taking in mind scalability issues. Finally, it presents the problems that arose when this network was put in place by using both commercial and experimental routers. I. INTRODUCTION Recently, the Internet has seen a growing interest in Quality of Service (QoS) issues. This is due to the increase of the traffic in the network and to the different services that are going to be provided through it. Particularly, commercial interests generated the need to differentiate the service on the application basis (for example bulk data transfers against real-time data) and on the customer basis (on-line traders are more sensitive to quality than home web-surfers). Up to now, the Internet Engineering Task Force (IETF) proposed two models to provide QoS guarantees on the Internet. The Integrated Services [1] (IntServ) model allows for a per-flow reservation and it is able to provide “deterministic” guarantees with its Guaranteed Services [3] specification. That is, a user can book in advance the amount of bandwidth it needs on a specified path and it can determine the end-to-end delay. The network will check for resource availability and it will accept the booking only if the request can be satisfied. A second specification, the Controlled Load [2], defines a service that corresponds to the one experimented on an unl

There are no comments yet on this publication. Be the first to share your thoughts.