Today, there are hundreds of production Apache HBase clusters running either entity-centric or event-based applications. While applications are a near perfect fit, others attempted applications are non-optimal, and several are somewhere in between.
Gathered from known clusters and a survey conducted by Cloudera’s development, product, and services teams and its the nearly 20,000 HBase nodes under management, this talk categorizes these the gamut of use-case into a compact set of application archetypes. We’ll help you with some simple questions you can use to characterize your application and determine if HBase-alone, if HDFS-alone or if some combination of the two the correct solution.
We’ll also explain techniques that an HBase application builder can use to implement these archetypes, how to handle multiple workloads on the same data, and discuss new features that alleviate pressure points of existing poorly fitting archetypes. Finally, we’ll use these results will provide insight into what lies in HBase’s future.
Software Engineer @ Cloudera. Apache HBase Commiter, Apache Flume Founder.
Author of O’Reilly’s “HBase – The Definitive Guide”.
Comments on this page are now closed.