Skip to main content

1990 | Buch

Computer Networks and Systems: Queueing Theory and Performance Evaluation

verfasst von: Thomas G. Robertazzi

Verlag: Springer US

Buchreihe : Telecommunication Networks and Computer Systems

insite
SUCHEN

Über dieses Buch

Statistical performance evaluation has assumed an increasing amount of importance as we seek to design more and more sophisticated communi­ cation and information processing systems. The ability to predict a pro­ posed system's performance without actually having to construct it is an extremely cost effective design tool. This book is meant to be a first year graduate level introduction to the field of statistical performance evaluation. As such, it covers queueing theory (chapters 1-4) and stochastic Petri networks (chapter 5). There is a short appendix at the end of the book which reviews basic probability theory. At Stony Brook, this material would be covered in the second half of a two course sequence (the first half is a computer networks course using a text such as Schwartz's Telecommunications Networks). Students seem to be encouraged to pursue the analytical material of this book if they first have some idea of the potential applications. I am grateful to B.L. Bodnar, J. Blake, J.S. Emer, M. Garrett, W. Hagen, Y.C. Jenq, M. Karol, J.F. Kurose, S.-Q. Li, A.C. Liu, J. McKenna, H.T. Mouftah and W.G. Nichols, I.Y. Wang, the IEEE and Digital Equip­ ment Corporation for allowing previously published material to appear in this book.

Inhaltsverzeichnis

Frontmatter
Chapter 1. The Queueing Paradigm
Abstract
This is a book about statistical prediction. It tells the story of how the behavior of complex electronic systems can be predicted using pencil, paper, the poetry of mathematics and the number crunching ability of computers.
Thomas G. Robertazzi
Chapter 2. Single Queueing Systems
Abstract
In this chapter we will look at the case of the single queue. Even though chapters 3 and 4 will discuss networks of queues, it is probably safe to say that the majority of modeling work that has been done for queues has involved a single queue. One reason is that for any investigation into queueing theory the single queueing system is a natural starting point. Another reason is tractability: much more can be easily ascertained about single queues than about networks of queues.
Thomas G. Robertazzi
Chapter 3. Networks of Queues
Abstract
In this chapter we will consider networks of queues. Simple analytical results are usually only possible for Markovian queueing networks. We will start by establishing the product form solution for the equilibrium state probabilities for such networks in section 3.2. The existence of the product form solution basically means that the joint state probability can be expressed as a simple product of functions associated with a network’s individual queues. In the case of open queueing networks of state independent queues these functions are simply the marginal state probabilities so that it seems that the queues act as if they were independent. This interesting observation was first noted by J.R. Jackson [JACK 57] in the original product form paper for open networks in 1957 and later generalized in [JACK 64]. Later, in 1967, W.J. Gordon and G.F. Newell [GORD] demonstrated the existence of the product form solution for closed networks. In 1975 F. Baskett, K.M. Chandy, R.R. Muntz and F.G. Palacios [BASK 75] generalized the families of queueing networks known to have the product form solution.
Thomas G. Robertazzi
Chapter 4. Numerical Solution of Models
Abstract
The ultimate goal of much performance analysis is to generate numerical performance results for a particular system under specific conditions. These can take the form of tables or a set of performance curves. They are a variety of numerical techniques available, ranging from the brute force solution of the state equations to clever algorithms to Monte Carlo type simulation. With all these techniques the cost of solution in terms of solution time and computer memory requirements are important considerations. The cost of of solving a system of even moderate complexity may be prohibitively expensive for a particular technique and computer installation. There is a tradeoff between our ability to model and our ability to solve such models. There is thus an advantage to simplified models which capture the most important aspects of the system in question.
Thomas G. Robertazzi
Chapter 5. Stochastic Petri Nets
Abstract
Petri Nets provide a means for modeling and graphically representing the possible behavior of systems in which concurrency, serializability, synchronization and resource sharing are important considerations [PETE], [FILM]. They can be used for the understanding and prediction of the behavior of computer systems, communication protocols, biological, economic and other complex systems.
Thomas G. Robertazzi
Backmatter
Metadaten
Titel
Computer Networks and Systems: Queueing Theory and Performance Evaluation
verfasst von
Thomas G. Robertazzi
Copyright-Jahr
1990
Verlag
Springer US
Electronic ISBN
978-1-4684-0385-5
Print ISBN
978-1-4684-0387-9
DOI
https://doi.org/10.1007/978-1-4684-0385-5