If this book had ended at chapter 4, I probably would have sent it back.

It starts out with extremely low-level detail, and takes a deep dive into the mathematics of systems and operations research, best suited for the hardcore systems and operations theorist rather than a reader who is simply interested in novel approaches to autonomic computing. That’s not to say that the theory isn’t novel; on the contrary, it seems to be quite novel. What it is to say, though, is that the mind-numbing, excruciating number of equations, theorems, and algorithms without so much as a breather or simple high-level example in the early chapters may turn readers away before they’ve had a chance to get to the more user-oriented and experience-oriented chapters, which begin at chapter 5.

Chapter 1 introduces a basic theory of analysis/optimization for multi-server systems. This core theory is a basis on which the remainder of the book can build, and includes ample theory and algorithms describing optimizations required in the multi-server age (another reason that self-*, or self-managing, systems came to be, according to the author). Chapter 2 builds on chapter 1, identifies how to select models for self-* systems, and provides a few use cases and examples based on the authors’ experience that demonstrate the utility of their selection approach. Chapter 3 explores the tradeoffs between estimation of self-* analysis models, describing the difference between statically estimating model-tuning parameters (more scalable and efficient, and thus appropriate for real-time decisions, yet not as accurate), and dynamically estimating model-tuning parameters (less efficient and scalable, but highly accurate and thus not appropriate for real-time decisions). Chapter 4 introduces quality of service (QoS) and its relationship to self-* systems, focusing on improving energy consumption as a QoS concern. Chapter 4 concludes the theory section, as well as the first part of the book.

The second part of the book addresses the higher-level and more user-oriented aspects of self-* systems. Chapter 5 introduces the selflet architecture, describing a new form of self-aware interface called a selflet (similar to a self-* aware Java servlet from the hypertext transfer protocol (HTTP) domain). The authors describe how selflets have the ability to dynamically provide services to other selflets via learning, direct provision, or redirection to another selflet that provides the service. Chapter 6 returns to runtime models, this time focusing on the models in the context of the PLASTIC project, a European research project focused on self-* systems. Chapter 7 rounds out the book with an interesting discussion of the self-* properties of Apache Hadoop.

As a reader, I found the latter three chapters (and second half of the book) interesting; the number of pictures, relevant examples, charts, and graphs were more than enough to keep my attention. The initial four chapters of the book were a bit too theoretical to keep my attention, and I’d wager that any casual reader from the systems and operations domain and outside the realm of systems and operations research would feel the same way.