Two of their best insights for me:
* Architecture affects the organization of the company/business unit. (In my company, we didn't realize this and we failed to create an organization that could support the architecture.)
* Virtually any architecture can accomplish the functional needs of a system - what differentiates architectures are how they provide the essential qualities (performance, modifiability, maintainibility, etc.) to the product.
The book is strongly based in the real-world, with practical examples. I never felt they were straying into "theorectical" land.
I also bought "Applied Software Architecture" but didn't like it nearly as much - I highly recommend "Software Architecture in Practice"!
The content is relevant, clearly described and trustworthy. It has very little references to alternative views on the subject.
I am still looking for a good introductory fat book on software architecture.
UML is now as fundamental a piece of knowledge to the software architect as schematics are to the electrical engineer, and for the same reason - it is a common language of discourse, and is supported by the available tools. I urge the authors to publish a second edition with UML substituted for the ad-hoc diagrams.
What I didn't like about the book was that it doesn't present a complete approach to developing a software architecture. It outlines many different pieces of the puzzle of architecting, but doesn't do a good job of putting the complete puzzle together for the reader.