Imagine a router in a data network. Packets arrive at random times. The router has a buffer that can hold 10 packets. The number of packets in the buffer at any moment is a Markov chain (given the current number, the past arrival pattern doesn’t help predict the next step). Stewart shows you how to write down the transition probabilities, find the steady-state distribution, and compute the probability of dropping a packet when the buffer overflows.
That feeling—the strange, frustrating dance of randomness, service, and waiting—is the domain of performance modeling. And if there’s one book that unlocks its mathematical soul, it’s William J. Stewart’s (2009, hardcover). Imagine a router in a data network
The exercises are excellent—theoretical derivations, computational problems, and open-ended modeling challenges. Many problems explicitly ask you to implement a simulation in a language of your choice (pseudocode is given, but the ideas translate to Python, R, MATLAB, or Julia). You might wonder: why not a newer book? Some topics (like cloud computing or modern load balancing) aren’t covered, but the fundamentals haven’t aged a day. Stewart’s clarity, structure, and mathematical care remain unmatched. The hardcover binding is also a pleasure—this is a book you’ll keep open on your desk for years, flipping between the Markov chain chapter and the simulation appendix. The number of packets in the buffer at
Many modern texts oversimplify or skip the Markov chain theory, jumping straight to simulation scripts. Stewart refuses to compromise. He knows that if you don’t understand the steady-state equations of a Markov chain, you won’t truly understand why your simulation output sometimes oscillates or fails to converge. No book is perfect. Stewart’s coverage of non-Markovian queues (like G/G/1) is light—he points to approximations (Kingman’s formula, Whitt’s QNA) but doesn’t develop them deeply. Also, the simulation code examples are in a pseudo-language that some readers might find dated; you’ll need to translate to your preferred language. But these are minor quibbles. The Takeaway William J. Stewart’s Probability, Markov Chains, Queues, and Simulation is not just a textbook. It’s a key to seeing the world differently. After you read it, a checkout line is no longer an annoyance—it’s a continuous-time Markov chain with finite waiting room. A busy website is a Jackson network of queues. Your email inbox is a discrete-time queue with a priority scheduler. And if there’s one book that unlocks its
We’ve all been there. You’re at the supermarket, holding a single item, staring at a dozen checkout lanes. You pick the shortest one. Naturally, it stops moving. The person in front of you writes a check. Slowly. A machine needs a price check. You glance at the next lane—it’s flowing like water. You sigh.
