Evaluating Software Architectures: Methods and Case Studies [Hardcover]

The foundation of any software system is its architecture, that is, the way the software is constructed from separately developed components and the ways in which those components interact and relate to each other. If the system is going to be built by more than one person—and these days, what system isn't?—it is the architecture that lets them communicate and negotiate work assignments. If the requirements include goals for performance, security, reliability, or maintainability, then architecture is the design artifact that first expresses how the system will be built to achieve those goals. The architecture determines the structure of the development project. It is the basis for organizing the documentation. It is the first document given to new project members, and the first place a maintenance organization begins its work. Schedules, budgets, and workplans all revolve around it. And the senior, most talented designers are paid to create it.

A system's longevity—how viable it remains in the face of evolutionary pressure—is determined primarily by its architecture. Some architectures go on to become generic and adopted by the development community at large: three-tier client-server, layered, and pipe-and-filter architectures are well known beyond the scope of any single system. Today, organizations are recognizing the importance and value of architectures in helping them to meet corporate enterprise goals. An architecture can give an enterprise a competitive advantage and can be banked like any other capitalized asset.

The right architecture is the first step to success. The wrong architecture will lead to calamity. This leads to an important question: If your organization is betting its future—or at least a portion of it—on an architecture for a system or family of related systems, how can you be sure that you're building from the right architecture and not the wrong one?

The practice of creating an architecture is maturing. We can identify causal connections between design decisions made in the architecture and the qualities and properties that result downstream in the system or systems that follow from it. This means that it is possible to evaluate an architecture, to analyze architectural decisions, in the context of the goals and requirements that are levied on systems that will be built from it.

And yet even though architecture is regarded as an essential part of modern system development, architecture evaluation is almost never included as a standard part of any development process. We believe it should be, and this book is an attempt to help people fill that gap.

The time has come for architecture evaluation to become an accepted engineering practice for two reasons. First, architecture represents an enormous risk in a development project. As we've said, the wrong one leads to disaster. It makes good sense to perform an evaluation on such a pivotal artifact, just as you would plan risk-mitigation strategies for other sources of uncertainty. Second, architecture evaluation can be remarkably inexpensive. The methods described in this book add no more than a week to the project schedule, and some abridged forms require no more than a day or two. Architecture evaluation represents a very cheap insurance policy. Compared to the cost of a poor architecture, the modest expense of a software architecture evaluation makes all the sense in the world. What has been lacking up to this point is a practical method for carrying it out, which is where this book comes in.

This is a guidebook for practitioners (or those who wish to become practitioners) of architecture evaluation. We supply conceptual background where necessary, but the intent of the work is to provide step-by-step guidance in the practice of architecture evaluation and analysis. To help put the methods into practice, we have included sample artifacts that are put into play during an architecture evaluation: viewgraph presentation outlines, scenarios, after-action surveys, final report templates, and so forth. The goal is that after reading this book, you will feel confident enough to try out the methods on an architecture in your own organization. We have tried to help answer the question, during an evaluation, "What should I do now?"

While the book is written from the point of view of the evaluator, there are others involved in an evaluation—project managers, architects, other stakeholders—who will gain valuable insights by reading this book. They will come to understand how their products will be evaluated and thus can position themselves to make those products fare better with respect to the evaluation criteria. This is rather like scoring well on a test because you've seen an early copy of the test, but in this case it isn't cheating but rather sound management and engineering practice. But know that when we use the word you in the text, we are speaking to the evaluator.

The techniques in this book are based on actual practice in government and industry. Most of the methods were developed by ourselves and others at the Software Engineering Institute and applied by ourselves and others to our customers' and collaborators' systems. Other material was gleaned by holding industrial workshops whose participants were experts in the analysis and evaluation of architecture. In short, we have learned by doing, and we have learned from others' doing.

This book will not teach you how to become a good architect, nor does it help you become fluent in the issues of architecture. We assume that you already have a good grasp of architectural concepts that comes from practical experience. This book will not help you assess the job performance of any individual architect nor a project's architecture (or development) process. What it will do is show you how to evaluate an architecture with respect to a broad spectrum of important quality attributes having to do with the architecture and the future system(s) that will be built from it.

Finally, we should say a word about software versus system architecture—that is, the architecture of software-intensive systems. This is a book about the evaluation of software architectures, but we often hear the question, "Well, what about the architecture of the system, not just the software? It's just as vital." We couldn't agree more. System architectures embody the same kinds of structuring and decomposition decisions that drive software architectures. Moreover, they include hardware/software tradeoffs as well as the selection of computing and communication equipment, all of which are completely beyond the realm of software architecture. System architectures hold the key to success or failure of a system every bit as much as the software architecture does for the software. Hence, they deserve to be evaluated every bit as much and for exactly the same reasons.

The methods presented in this book will, we believe, apply equally well to system architectures as to software architectures. If modifiability is a concern, the methods can be used to gauge the expense of making changes over the system's lifetime; if performance is a concern, the methods can be used to spot bottlenecks and problem areas in the system as well as the software; and so forth.

Why, then, do we call it a book about software architecture evaluation? Because that is the realm in which the methods were invented, developed, tested, and matured. In the remainder of this book when we speak of architecture, you can always safely prefix it with software. You can prefix it with system depending on how applicable you feel the methods are to system architectures and how confident you are about our intuition in the matter.

As a final word, we invite you to share your experiences with us. We would be keenly interested in knowing what you discover works well and what doesn't work so well. Writing a book is an opportunity to share lessons, but more importantly to us, it is an opportunity to gather new ones.

—PCC, Austin, Texas—RK, Pittsburgh, Pennsylvania—MHK, Pittsburgh, Pennsylvania