Software Project Management: A Unified Framework [Hardcover]

Conventional software engineering is first presented, along with its drawbacks. The changing software landscape is examined, and future success factors are identified. The factors are then tied into the new process.

Thorough coverage of many different angles, including an excellent section on metrics, estimation models (and the evolution of COCOMO), team development, culture changes, etc.

Great read.

That being said, the "process framework" hinges on iterative development. The author does a good job of showing why the traditional waterfall process does not work in today's environment, and this is applies to all organizations.

Then the author goes on to outline how software development breaks down into phases and iterations within phases. He also talks about the activites and artifacts that belong in each phase. Additionally, he talks about how organizations should be structured to support the process and how to taylor the process to a specific organization.

The case studies support the claims made throughout the book making it well rounded and logically sound. This process will work very well in the correct situation.

The highlights are:
* A project life cycle and process framework that is [obviously] closely aligned to the Rational Unified Process (RUP), and can be fitted to any rapid development or iterative approach.

* An excellent tutorial on effective project controls, with an emphasis on earned value project management.

* In-depth coverage of estimating techniques, with a lot of material on the constructive cost model (CoCoMo), and current gaps in estimating techniques and to where the craft and science of estimating and software economics needs to evolve in the discussion of next-generation cost models. I especially like his distinction between the use of source lines of code metrics for size and function points for scale. There is middle ground.The treasure trove of metrics, including core project metrics, and the change metrics that are given in Appendix C.
There is one glaring flaw in this book and an experienced project manager will quickly spot it: the proposed approach to basing work breakdown structures on project phases instead of the decomposition of the system to be delivered will not work. Using Royce's approach there is no clear way of integrating the work breakdown structure with the organizational breakdown structure. Using earned value techniques (which is well covered elsewhere in the book) Royce's approach will not align control accounts (sometimes called cost accounts), making his recommendations contrived and unworkable.

This book is better suited for an architecture-centric approach to project management, which means that it's more applicable to product development instead of internal IT projects. However, all seasoned PMs will learn much from it.