J.D. Meier recently posted a summary / impact statement on the Microsoft Blue Books, and I thought I'd add a thought from my experience.
Recently I was engaged with a customer who was considering updating their solution stack. They had decided to update their database servers to leverage some of the new SQL Server 2008 features, and they thought it might be a good idea to modernize their solution along with the effort.
This was a great opportunity for them, but at the same time there was a lot of work in understanding the range of options available to them and what tradeoffs came with each option.
The decisions that you make when you lay out the fundamental pieces of your architecture last a very long time, and typically cascade into a multitude of other decisions down the line. For example, if I choose to use custom serialization across a component boundary, the connected pieces better be able to consume and produce this format, which already lays out a constrained set of options for me.
Taking the time to explore options and work through the different trade-offs explicitly is therefore a high "return-on-investment" activity if done right. One definitely documenting, by the way 🙂
So start from the beginning with the great assistance the Blue Books give you, and stand on the shoulders of those who have come before and learned some hard lessons.