My general review of this book was that inevitably, software will reduce the amount of people needed to write--you guessed it--software. I got the general trend that using a cloud-based solution was optimal for providing quick turn-around, an already proven environment, and costs that scale as your needs do.
I disagree in general with cloud-based solutions at this time, however, as I don't think they've proven themselves as anything beyond specialized "walled gardens" of data. The simple fact that you cannot EASILY export and import your data from one cloud solution to another is a sign that this market isn't broad enough yet to require a standard.
- File systems have standards - File Types have standards (XML, CSV, YAML, HL7, X.509, etc.) - Protocols have standards (SSH, HTTP, FTP, etc.)
But for the most part, every cloud solution 'rolls their own' way of doing everything, resulting in something that might be extremely flexible for their architecture, but which ultimately leaves people who use their tools in a bit of a bind.
Overall, I enjoyed the read, but I don't think I'll be picking up their suggestions. Becoming an expert in one cloud-based tool, when there are additional ones out there, means that I won't be useful outside of a niche domain. Google Apps, Amazon EC2/S3/etc., MS Azure, and a whole host of smaller players--all provide a somewhat-similar, but not standard solution.