It's written as more of a comprehensive guide, walking you through concepts and talking about them in a bit more depth than the docs. I also found the code examples to be much more helpful than what was in the docs. I believe it also goes more in depth (guide-wise) on many topics than the docs do.
In my experience as you are learning to use ElasticSearch during an implementation, or even going back to reference there is a high-ish probability you will need the manual and the definitive guide. Really, I think they ought to be combined in some way because there are huge gaps in the docs.
I bought the book and I'll say I was disappointed with the book. A couple of points, though. 1. The topic is huge, it's hard to cover everything because the way you can query and setup your mappings is really very open-ended. 2. The book is a good way to get started, but it lacks as a reference. See point 1., a reference may not be possible.
The best way to learn ES is to set yourself up with a scripting language (python, ruby, etc.) and prototype everything, then apply what you learned to your language of chocie (i.e. Java if the scripting language isn't your language of choice).
TBH, I'd prefer a manual. In fact, I'd prefer frequent use case examples, documentation and flow charts for more in depth understanding. English takes too much time for me to scroll through : /
But I can understand the appeal. If you're looking to feel more intimate with the technology from another persons perspective or do not want terrible documentation, this kind of style makes sense I suppose
