Resource allocation and load balancing mechanisms for providing quality of service in the internet

Abstract

This work introduces mathematical models and algorithms for resource allocation and load balancing in various network elements. The underlying principle on which the technical focus of this work is targeted is the Quality of Service (QoS) mechanisms at different layers of networking - the work deals with the protocol layers from optical transport networks to Web applications' load balancing. The algorithms and models are mainly studied by simulations. Among the resource allocation strategies, the first part of the work covers dimensioning of a high-speed switching bus, Frame Synchronized Ring (FSR), that is used in Software Configurable Multidiscipline Switch (SCOMS). The rest of the work is dedicated to a load balancing algorithm and its variations to network server load balancing, to traffic flow balancing in Multiprotocol Label Switching (MPLS) and Wavelength-Division Multiplexing (WDM) networks, and to an application for process control networks using Fault-Tolerant Ethernet (FTE). In MPLS networks, the use of the algorithm is integrated into pricing mechanisms. This work can be reasonably seen as a series of temporary answers to a number of open questions in the field of QoS. The eld evolves very fast, and there are not too many technological solutions that can be foreseen to remain fixed for a long time.