A few weeks ago Packt Publishing sent me a free copy of their new publication - Getting Started with Hazelcast by Mat Johns to read and write about. I have used distributed caches and compute grids quite a bit at work. So, I was happy to do a quick review of this book. I've used Oracle Coherence quite a lot and Hazelcast for some experiments.
The book is a gentle guide to building distributed compute and data grids. It assumes nothing about the reader and hence does a good job of doing what it says in the book's title - "getting started". I'd advice this book for anyone who is completely new to this area which is not to be confused with Hadoop, Storm, Cassandra or the other more "popular/hyped" cousins. I would say that for medium sized data, logic heavy, transactional/near real time applications, compute grids are the way to scale out.
Obviously this book is about using Hazelcast, which is a nice Apache software licensed, Java, distributed grid/cache. It is surprisingly feature rich and in terms of usability, features and elegance it comes very close to its more expensive, older, rock solid cousin which is Oracle Coherence.
The book explores the essential aspects of using such frameworks effectively. Such as - distributed maps, replication, network partitions, fault tolerance, data affinity, moving code closer to where data is etc. It does this without being too overwhelming for first timers.
For a full and more thorough treatment I would obviously recommend the Hazelcast documentation. And if you are curious to know about other frameworks check out my old write up - Scalable compute & storage frameworks - A Refcard.