This book spent a year on my desk. It had been recommended by two of my friends by it was intimidating enough to always find something more important or fun to read. 4 days ago I have decided to finally get rid of it from my desk :)
Turns out, most of the things that were interesting I have already knew from M101JS: MongoDB for Node.js Developers. I mean 'sharding' and 'replication' are the most fundamental knowledge about all databases so it was worth to read about them again, but still, nothing new. Fortunately there were also new ideas. First, CAP Theorem that was nicely explained and I finally understood the importance of setting how many clusters need to acknowledge the reads and writes. Second good thing was reading more about Map-Reduce in context of clustered databases.
Book is really short and is much easier to read than you might think.