O'Reilly Learning Spark: Lightning-Fast Big Data Analysis

The writing itself is good, relatively concise and uses simple, almost conversational language to describe Spark. The focus is on using Spark, at an Entry level, with examples covering the three main Spark languages: Scala, Python and Java.The content is clearly outdated, with RDDs used as the most frequent data abstraction and Data frames/Datasets left as an afterthought. There are some gems in here that you won't find explicitly mentioned online, but it isn't intended to be a deep-dive or best practices book.I'd recommend this book for people interested in understanding the fundamentals that Spark is built upon, and how current Spark releases have developed since 2015.However, I would generally recommend waiting for the soon-to-be-released 'Spark: The definitive guide', which will provide a much more thorough and up to date guide. If you want more specific knowledge about spark internals (I would recommend that any spark user should), best practices and optimisations then buy 'High Performance Spark' also by Holden Karau instead of this book.

I think Spark is going to be a tough subject to get your head around so you have to expect that there'll be no book that's easy reading. Having said that, this book was pretty good for introducing the subject. I began reading and working through the sample code at the same time but found this too time consuming so decided to read it all and I'll come back to the different examples as I need them. That worked best for me. Whilst I'm still no Spark developer, I do feel I have an understanding of what it is and how it works. This means when I go to develop anything I have a rough idea where to begin.

This is a good introduction to Spark, it doesn't attempt to be a detailed deep-dive into the internals. The overall pace of the book is fine. My one criticism is that the final chapter on Machine Learning seemed a bit rushed and would have benefited from a clearer introduction to the topic and a more detailed walk through a few examples. The GraphX library - which is a very interesting part of Spark - doesn't have a chapter which is a shame. Overall good, but in a 2nd edition, I would hope the MLib section gets a re-write and GraphX has its own chapter.

1. Don't buy the paperback verson as the blurb on the Amazon site says " Recently updated for Spark 1.3" but the paperback isn't as I found out when I received it.2. Most of the material in this book is available online at the Apache Spark website.It's not a bad book, but not sure I can recommend it as it doesn't add much beyond what is freely available.

You can learn Spark without this book in my opinion, but if you like learning from books this one will give you basic skills.