Software Architecture

BShive writes "Software Architecture: Organizational Principles and Patterns covers the VRAPS model and the organizational aspects of Software Architecture. Patterns and Antipatterns are explored that resolve or complicate problems depending on the criteria involved. A Pattern that solves one situation might become an Antipattern in another, as not all situations need the same solutions. This fact is something forgotten too often in software projects. Architects, coders and even managers might benefit from the information contained in this book. Being able to identify and solve problems in a project and its organization is important for any large software project no matter where you are in the development chain." Read on for the rest of Ben's review. Software Architecture: Organizational Principles and Patterns author David M. Dikel, et al pages 250 publisher Prentice Hall PTR rating 7 reviewer Ben Shive ISBN 0130290327 summary Useful approach to organizing software projects, from people to code.

The book opens by explaining what VRAPS (Vision, Rhythm, Anticipation, Partnering, and Simplification) is and what the book can do for the reader. Software Architecture is increasingly important, but the organizational aspect is often overlooked. Architecture and Organization do overlap, but to the executive the Architecture side is hidden, and to the practitioner the Organizational side is hidden. VRAPS attempts to shift the perspectives of the executive and practitioner to provide a more balanced view. An excellent summary of why each of the VRAPS principles are important is provided. A short example scenario follows, briefly illustrating how the model can be used and misused. These concepts are further expanded throughout the book.

The second chapter is essentially a more detailed look at VRAPS and how everything fits together. Criteria, Patterns and Antipatterns are explained, along with a short history of VRAPS. An amusing anecdote mentioned was a manager who divided his program into one hundred modules to show percent complete. Only five modules had more than 100 lines of code. One of the five had over a million lines. There are similar occurrences throughout the book that illustrate various follies in software development and management.

Chapter three deals with maintaining the vision and direction of the project while balancing all the influences. To a manager, the project may look perfectly ordered on paper while features are added and removed. On paper it still looks neat, but to the practitioner it can appear a jumbled mess. The reader also sees the first example of how the situation layouts are handled in the book. A short summary covering the Criteria, Antipatterns, and Patterns is presented. Then each criterion is further examined with its related Antipatterns and Patterns.

Further chapters proceed with introducing various development concepts that complete the VRAPS moniker. How to put the concepts into practice is explored through the same Criteria, Antipattern and Pattern layout. It does an excellent job of illustrating each part of VRAPS. Following at least some of the principles will result in a project that will be successful, instead of becoming one of the book's examples where the team ended up with nothing to show for its work.

The chapter on the Allaire (now part of Macromedia) case study was the most interesting chapter of the whole book. Company and product development is followed, including mistakes made along the way. The final chapter on 'Building and Implementing a Benchmark' was rather unimpressive. It seemed merely tacked onto the end and included no real conclusion to the entire book. However, the rest of the book is a solid piece of work with very useful information.

The anecdotes and examples throughout keep the reading from becoming too dull. Even with a flat finish to the book it contains plenty of valuable information and is worth the admission price, though it could have been better still.

Chapters

1. What You Can't See could Help You
2. The VRAPS Reference Model: How the Pieces Fit Together
3. Projecting and Unifying Vision
4. Rhythm: Assuring Beat, Process, and Movement
5. Anticipation: Predicting, Validating, and Adapting
6. Partnering: Building Cooperative Organizations
7. Simplification: Clarifying and Minimizing
8. Principles at Work: The Allaire Case Study
9. Case Study: Building and Implementing a Benchmark Using VRAPS

Appendixes

A. Quick Reference Table: Principles, Criteria, Antipatterns, and Patterns
B. Antipattern and Pattern Summaries

You can purchase Software Architecture: Organizational Principles and Patterns from bn.com. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page.

Comments

At work...

[Anonymous Coward]

we use the American Standard Spiral (ASS) model.

From the department of the Glaringly Obvious

[L. VeGas]

Being able to identify and solve problems in a project and its organization is important for any large software project no matter where you are in the development chain.

This is the kind of astute observation that makes reading /. so fun (and informative!). If I may be so bold to add on to your analysis, I think it's also important for a software developer to be able to read and write. No matter where you are in the development chain.

Re:From the department of the Glaringly Obvious

[Tablizer]

[Being able to identify and solve problems in a project and its organization is important for any large software project no matter where you are in the development chain. ] This is the kind of astute observation that makes reading /. so fun (and informative!).

I once took some Management courses, and roughly 80 percent of the textbook was stuff like that. There was very little real substance. I suppose to people just born (physically or mentally) it might be new, but it bored the sh8t out of me.

At least I learned why PHB's talk the way the do, and perhaps learned some good fluff-talkin' techniques for speeches, interviews, etc. I suppose that counts as "education", but I doubt it is what the author(s) intended.

It's a go!

[Hayzeus]

This methodology has what is probably the most important feature any methodology can have: a nifty acronym. Sounds highly technical, yet it could also be a tasty lunch item. I shall write a memo to the CTO forthwith!

Re:It's a go!

[Mr. Sketch]

Actually the acronym is missing an MT in front of it for Multiple Threads, making the real acronym MTVRAPS.

Re:It's a go!

[ergo98]

Acronyms and titles on processes are often a great source of hilarity as well meaning and inferior feeling developers will go along with whatever you say just to seem like they're "in" with whatever is hip and cool (despite the fact that the overwhelming majority of these things are fringe technologies and processes that overwhelmingly people have no clue, rightly, about).

"Are you familiar with the CORAN 2 process?"
"Oh yeah...we use that a lot."
"Really? I use it in concert with UMX and ICBM VSLAM for maximum effect. We use Agile Extremities processes with core-duplex programming methodologies"
"Ooooh...sounds awesome!"
"Yeah, it's good stuff. You really need quad-programming to and read once write never methodologies to have quality code. As long as you use over the shoulder management with sycophant posterior gestulations it all turns out good."

manager?

[tanveer1979]

An amusing anecdote mentioned was a manager who divided his program into one hundred modules to show percent complete.

You don't call such people managers....
you call them damagers.

Saw presentation earlier this year

[mwillson]

I saw David Kane present these concepts at a Software Architecture Conference. In essence, he was reinforcing the importance of architecture, and the need to deal adequately with all the non-technical aspects of implementing an architecture within an organisation. The book is worth a read, mostly for solutions to people-type problems (e.g. how to persuade a development team to adhere to an architecture, using informal relationships to aid communication and shared goals).

VRAPS

[sql*kitten]

Software Architecture: Organizational Principles and Patterns covers the VRAPS model and the organizational aspects of Software Architecture

You know, there's one thing worse than developing software without a methodology, and that's changing the methodology every time someone comes out with a new acronym. No-one can evaluate a method until they've done a few non-trivial projects with it, and that takes years. If all the people who jumped on the RUP bandwagon then the XP bandwagon jump on this, the industry's track record for delivering on time and within budget will only get worse.

Re:VRAPS

[spakka]

The book under review seems even more vacuous than usual for the genre. The chapter titles are laughable, but appealing to a certain 'inspirational leadership' type of manager. Only one successful case study? Design Patterns cites several per pattern. And I simply don't believe the anecdote about the manager with the million+ lines of code in his source file.

Re:VRAPS

[pmz]

No-one can evaluate a method until they've done a few non-trivial projects with it, and that takes years. If all the people who jumped on the RUP bandwagon then the XP bandwagon jump on this, the industry's track record for delivering on time and within budget will only get worse.

Thus the importance of not adopting RUP, XP, etc. for real projects. These methodologies can be informative, but it is better to create a simplified custom process for each project. It isn't very hard, and the development team can establish the tool chain, conventions, and documentation methods that suits them and the project's requirements best. Note that simplifying the process is critical, because no one can seriously keep track of developing real software while trying to learn some baroque process. Also, it is always critical to avoid proprietary documentation formats (e.g., basically anything by Microsoft), trendy IDEs, acronyms of the month, and other neat but immature development toys.

Personally, I think taking the time to actually implement the dogma of RUP, XP, etc. is a waste of time, when 1) no one really understands them, anyway and 2) they are like fashion: here today, gone tomorrow, possibly reborn in 20 years, but who knows.

Better acronym:

[Idarubicin]

Although VRAPS does emphasize Vision, Rhythm, Anticipation, Partnering, and Simplification, the whole book is about Criteria, Antipatterns, and Patterns.

Consequently, I propose the following acronym.

Criteria
Rhythm
Antipatterns
Patterns

Software is ART.

[Boss, Pointy Haired]

I wish people would stop trying to manage it.

Re:Software is ART.

[Daleks]

If software is art, what's the software equivalent of modern art? My guess it's something like Befunge or Brainf*ck. Also, I say the software world's Bob Ross would be Donald Knuth.

Yes, it's Monday.

Anti-pattern Rant

[Shamanin]

I had a former collegue that just couldn't grasp the use of design patterns, and thus despised the concept. He also couldn't solve large scale programming problems and wasn't much of a software architect in general. Then, the book anti-patterns comes out which he latched onto as some sort of weapon against the evil design patterns. He became very dogmatic about using what he had learned from the book to shoot down others designs (obviously, not what the book was intended for), trying to find fault in others work to somehow cover for his own inadequacies.

My former collegue taught me one thing if nothing else. It is easier to find a problem in a design than it is to find a solution for a design. Design patterns are a powerful way to classify and grasp large abstract recurring design issues. Anti-patterns are nothing but the same. Don't let anyone tell you otherwise.

Re:Anti-pattern Rant

[Bazzargh]

This isn't just a pattern/antipattern thing. Many discussions er... have this pattern. Its the reason why we have the phrase "constructive criticism".

Another extreme of criticism is Bikeshed Painting, where you get swamped with alternative designs; I only mention this because of the nice story behind that link.

Re:Anti-pattern Rant

[rossifer]

I had a former collegue that just couldn't grasp the use of design patterns, and thus despised the concept. He also couldn't solve large scale programming problems and wasn't much of a software architect in general. Then, the book anti-patterns comes out which he latched onto as some sort of weapon against the evil design patterns.

Ya know, I'll bet he loved the "Golden Hammer" antipattern. For those in the cheap seats: the golden hammer antipattern observes that people who get a shiny new tool tend to look at all new problems as if the tool can solve them. I.e. if the only tool you've got is a hammer, all of your problems start to look like nails.

This particular application of this anti-pattern (as a universal pattern debunking argument) is particularly ironic.

Regards,
Ross

This brain rot gives me a headache.

[nadador]

This is exactly what's wrong with the universe, or at least the small part of the universe occupied by software engineers.

What has all of our Functional, Object Oriented, Extreme Programmed, UML-based, XML compliant, Pattern-ed or Anti-Pattern-ed flow charts in animated PowerPoint got us? Its got us a load of crap, thats what. A load of crap. We re-org endlessly. We have more meetings. We write more Standard Operating Procedures. We rewrite the coding standard. We switch languages, run times, operating systems, and libraries. We refactor, re-code, re-work, re-design and re-plan. And we get a load of crap. We manage, and plan and re-manage and re-plan, depending on what the winds of your upper management's whims dictate is the "in" style for the day.

What should all of this tell us?

Software engineering is a practical craft. No amount of process will ever make up for proper training, proper documentation, proper version control, and proper testing. Ever. And that's the way it is. If you have good people, set them free. If you don't, spend a little money to train them to their highest potential instead of trying to make them good cogs in a crappy buzzword wheel.

In the end, 99% of the work done by software engineers is just rearranging magnetic pixy dust on some drive platter, or scattering the electrons in a flash or DRAM or SRAM cell. Most of our value to the universe is just damned pixy dust. And it shouldn't be this difficult.

We don't need any more of this - we all just need to learn how to be practical craftsmen that get *work* done.