23 September 2009

Software architecture book reviews

Some comments on textbooks on software architecture:
I would say SAiP is the standard textbook on the subject in the sense it covers the what is commonly accepted as the main core of the subject of software architecture. It probably also is the most widely used text...
As a student text I think the main strengths is it's treatment of terminology and concepts, which also makes it useful as a reference.
A potential problem with this text is that the examples are vey high level. You need experience based in actual programming projects to see the connection between the examples in the book and actual implementations. It is also comparably weak in teaching how to design a top-level desing as part of an architecture. I would say this book is suitable for students at the end of their education.

A Software Architecture Primer is targeted directly to a student audience. I would say it is better if the knowledge in an architecture course should be used in later courses, e.g. a larger project course with co-operating teams, since it is more pragmatic and hands-on in defining a top-level design. Likewise I would guess it could suit a professinal who wants to start creating "simple architectures".

The Process of Software Architecting is the best (and may be the only one) that actually defines the process elements and tasks of architecting and how they would fit in a larger software development process. I would say this book is the best one for the professional architect that wants to more formally define or improve what they do; what inputs to use, what tasks to perform and what should be delivered. It may be suitable for students, but the process focus may make it harder to relate to, e.g. it would give very little help as a preparation for a student project course.

I like all three books and they all have their (different) purpose. Which book to get depends on who you are and what you intend to do or to learn. For a student in the first course on architecture I would suggest the Primer. As a S/W professional who wants to understand what an architect is talking about and why, go with SAiP. For a working architect who wants inspiration for improvement, go with the Process book.

Background: I got the opportunity to take a sabbatical semester from Volvo to teach an undergraduate course on software architecture and design of complex systems at the IT University in 2006. For various reasons I redesigned the entire course from scratch, with a few constraints; the teaching methodology should be a light version of problem-based learning (PBL) already established at the IT University, and the course book was already ordered by the university bookstore "Software Architecture in Practice" by Bass, Clements and Kazman.
I had no real problem with the choice of textbook, and it suited my intended course content quite well. I also thought the PBL methodology fitted very well with the topic of software archtiecture after the course.
I think the course was a success, it received praise form the students as one of the most interesting courses they had taken during their program, and I was very satisifed with the level of knowledge the students showed on the final exam.

16 September 2009

Presentation on the Joint Working IEEE/IFIP Conference on Software Architecture & European Conference on Software Architecture

Here are the slides I used at my presentation at the Joint Working IEEE/IFIP Conference on Software Architecture & European Conference on Software Architecture in Cambridge 2009.

The presentation went well, but the questions showed that the audience was more interested in how the organisation perceived the role of it's own software architecture rather than the methodology and scientific conclusions. And the role of architecture was one thing that was not possible to generalise to other organisations...

10 September 2009

Accurate numbers

I am going to present my paper at the Joint Working IEEE/IFIP Conference on Software Architecture & European Conference on Software Architecture on Wednesday. If I get a question where somebody wants me to quantify my qualitative research I need to refrain from doing this...
Dilbert.com