HBase in Action

Many HBase books present HBase as a checklist of APIs; this one does not. Both authors have spent years solving real problems with HBase in production—Nick Dimiduk was using HBase as early as 2008 to tame ETL jobs that had been taking over twenty hours, and Amandeep Khurana worked on solution architecture at the Amazon Elastic MapReduce team and later at Cloudera. As a result, the book is full of the trade-offs behind "why design it this way" rather than "call the API like this." The through-line is a single example application, TwitBase, which runs from table creation, reads and writes, and scans all the way to row-key design, denormalization, coprocessors, and multi-language clients via REST/Thrift, so readers can see a table evolve from a draft into a design that can withstand its access patterns. The two chapters in Part 3 on OpenTSDB and geospatial indexing are especially valuable: the former is a retrospective on time-series modeling in a real open-source project, and the latter breaks geohash spatial indexing down to a level you can implement yourself—content that was rare in Chinese books of the same period. The translator is Xie Lei, founder of the ChinaHadoop technical community, and the terminology and code comments are handled in a practical, down-to-earth way, making the book well suited to reading while typing along. One caveat: the book corresponds to an early version of HBase (it still uses the -ROOT- table), so details such as cluster sizing and region location differ from today's 2.x versions; separate the underlying principles from the version-specific details as you read.