Evaluation of Dynamic Routing Strategies in

This works studies the performance of dynamic routing in QoS networks. Basing on the simulation of realistic networks, the paper compares the behaviour of dynamic and static routing strategies when they are applied to transmit the packets of different traffic classes in a Diffserv environment. The study also shows the effects on dynamic routing of employing MPLS. I. INTRODUCTION The increase of traffic volume in Internet and the tendency to integrate real time services oblige IP networks to support Quality of Service (QoS). QoS must be provided to respect the strict delay and loss requirements of new IP services, such as voice or videoconferencing, but it can be also regarded as a differentiating factor for Internet Service providers (ISP). In order to enable (or improve) QoS in Internet, Internet Engineering Task Force (IETF) has proposed several technologies. The first proposal was IntServ (Integrated Services ) (1). This architecture guarantees the particular QoS requirements of each flow between any two terminals of the network by reserving specific resources (mainly bandwidth and buffer space) in the routers that the flow traverses. The main disadvantage of Intserv is its lack of scalability as long as all the nodes (including internal or 'core' routers) have to give a particular treatment to the packets of each flow demanding QoS. Thus, its utilisation is confined to small sized networks, not being adequate for the backbone of most ISPs. As an alternative other two compatible technologies have been considered: Multiprotocol Label Switching ( MPLS ) (2) and Differentiated Services (DiffServ) (3). MPLS does not introduce by itself any QoS policy but it implements a new technique for packet forwarding. MPLS routers identify and forward packets basing on a simple 4 byte label which is inserted before the IP header. This label, which is negotiated during the connection set-up, avoids the high time-consuming IP forwarding performed by Intserv routers which require to analyse several fields in the packet headers (e.g.: origin and destination IP addresses and TCP/UDP ports). Thus, MPLS solves a part of the scalability problems which result from Intserv. However, as in the case of IntServ, MPLS networks present a difficult deployment since its labelling procedure demands all the nodes in the network to migrate to MPLS simultaneously. A single router in the network which was unable to interpret MPLS labels would prevent the whole network to run MPLS communications.

[1]  Hossam S. Hassanein,et al.  Quality of service based end-to-end SiMO routing framework in differentiated services networks , 2002, Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie.

[2]  Mario Gerla,et al.  Cost-effective multiple QoS path provisioning , 2002, Global Telecommunications Conference, 2002. GLOBECOM '02. IEEE.

[3]  David L. Black,et al.  An Architecture for Differentiated Service , 1998 .

[4]  Klara Nahrstedt,et al.  Hop-by-hop routing algorithms for premium traffic , 2002, CCRV.

[5]  John Wroclawski,et al.  The Use of RSVP with IETF Integrated Services , 1997, RFC.

[6]  Giovanni Pau,et al.  Practical QoS network system with fault tolerance , 2003, Comput. Commun..

[7]  Hossam S. Hassanein,et al.  Quality of service based end-to-end SiMO routing framework in differentiated services networks , 2002, Conference Proceedings of the IEEE International Performance, Computing, and Communications Conference (Cat. No.02CH37326).

[8]  Ward Whitt,et al.  Characterizing Superposition Arrival Processes in Packet Multiplexers for Voice and Data , 1986, IEEE J. Sel. Areas Commun..