This document is divided into three logical parts, split across a number of chapters. The first part (Chapter 1 and Chapter 2) contains a definition of the problem, a general introduction of the status quo and an overview of the problems associated with this approach. This is followed (Chapter 3 to Chapter 7) by a detailed examination of the problem areas that were identified in the first part as well as the solutions employed to overcome them. The last part (Chapter 8 and Chapter 9) of this work draws these problem areas together to present an overview of the work covered, and draws conclusions on them.
This section is intended to give the reader a brief overview of the structure of this document and the composition of each chapter.
Chapter 1 contains an introduction to the problem area. It also looks at some of the conventions used in this document and introduces the reader to the Internet standards process and the various Open Systems Interconnect models.
Chapter 2 uses the Open Standards Interconnect model of network management to identify the goals of network management, and looks at these goals in relation to current network monitoring techniques. It goes on to examine some of the shortfalls of these current methods, paying particular attention to the problem of multi-vendor networks. The chapter concludes with an introduction to the network at Rhodes University, which is used as a case study throughout this document.
Chapter 3 starts off by looking at simple network monitoring techniques aimed at the networking layman. The problem of multi-vendor networks is revisited and examined in detail. A solution to this issue is proposed, using the extensible markup language (XML) to provide a way of abstracting from the vendor-dependent components. An application that monitors several digital subscriber lines is developed to test this abstraction approach.
Chapter 4 looks at two different methods of determining topographical information about a network. The first of these, a traceroute(8)-based approach, produces network maps at layer three of the Open Systems Interconnect reference model, while the second, a simple network management protocol-based approach, produces maps at layer two.
Chapter 5 examines an intrusive way of experimentally determining the growth rate of a particular network. It examines the problems encountered in implementing this approach, as well as some of the social considerations associated with this sort of network monitoring. The problem of representing the vast amount of data gathered by this system is examined, and some unexpected uses for the information are discovered and discussed.
Chapter 6 provides a means by which network administrators can determine the location, both logical and physical, of a particular host on the network. It explains why this information is important, and discusses a number of problems related to obtaining it. A combined approach that attempts to work out both logical and physical location is developed and discussed in detail.
Chapter 7 explains that currently most network monitoring tools produce symptom-based fault reports. Several types of artificial intelligence are examined in order to find way of producing more useful fault reports, and an expert system is developed to do this. The result is an "intelligent" method of gathering information about the status of a network and of reporting faults.
Chapter 8 takes a broad overview of the work covered and explains how each of the systems presented are related to each other, and how they depend on other systems. Each application is examined in relation to the Open Systems Interconnect network management model in order to show how each system relates to the conceptual areas in this model. In addition, the problem of porting each of the systems to other networks is considered, with specific attention to those applications which are not straightforward to port. A comparison to existing solutions is also made.
Chapter 9 concludes the work and presents some possibilities for future work in this field.