Hadoop The Definitive Guide 3rd Edition Review
O’Reilly Hadoop 3rd Edition
June 2012 – O’Reilly publishers has released the 3rd edition of Hadoop The Definitive Guide, by Tom White. The author who is at Cloudera and created the first two editions, is been involved with the Apache Hadoop committee since Feb of 2007. His deep understanding a long time on the topic allows the book to be both technically accurate and in-depth, but more importantly surprisingly readable and understandable by those who are unfamiliar with the topic, and need to learn as well as those with high level of expertise and experience in the topic and need a complete reference guide.
The third edition includes all of the content from the first and second editions, and had been updated for the current version 2.x (0.23 release) of the software. The main areas of addition are the discussion on the newly added MapReduce API, and how to migrate from the old API, as well as MapReduce2 and its associated resource management system YARN. The book covers migration from the original MapReduce API to the new YARN environment only for MapReduce2. The book also reviews additions and changes to the HDFS and WebHDFS file systems. Finally the ripple effect updates on the Pig, Hive, Sqoop, and ZooKeeper to support new features and interfaces of the 2.x release are included.
For those considering a big data storage solution, this book is essential, even if you do use a Hadoop environment. It clearly outlines good and bad fit data and applications that are supported by Hadoop. As a distributed file system with inherent redundancy, it is not best suited to large contiguous data files that cannot be split and reconstructed easily. Understanding these tradeoffs helps determine where in a workflow to best use the technology, and also for determining which cloud service provider and storage solution is applicable for your storage requirements. If you are not an administrator, a number of the chapters are not applicable. The initial chapters however put the guide in the category of being a required reference for anyone directly involved with data creation in a workflow and how big data storage works today.
