Êtes-vous un étudiant de l'EPFL à la recherche d'un projet de semestre?
Travaillez avec nous sur des projets en science des données et en visualisation, et déployez votre projet sous forme d'application sur Graph Search.
We analyse and illustrate two approaches that we have developed for providing reservations to flows. The first approach uses the explicit reservation model: a traffic profile is negotiated between users and the network. ATM and RSVP are two examples for this approach. In this paper we describe our experience in the ACT project EXPERT (AC094) with the explicit reservation model in the context of ATM. We show that the integration of ATM in a TCP/IP oriented operating system is easy and does not require developing applications specially for ATM. The method we have designed is called Arequipa and is described in RFC 2170. It is fully implemented in Linux. However, we are confronted with the complexity required to support a single reservation for each flow. Our analysis is that it is well suited for segregating flows in applications such as private virtual networks, but that, for multimedia networks, the cost of handling a large number of flows is too high. This and other factors led us to develop an alternative solution, based on an implicit reservation model. Our solution is called the Scalable Reservation Protocol (SRP). SRP aggregates flows inside the network: routers other than edge routers performing policing do maintain only aggregate information per port. SRP is being developed and implemented in the framework of the ACTS project DIANA (AC319). Keywords: QoS; Aggregation; Arequipa; ATM; IP; SRP