Presentation: TweetApache HBase, the scaling machine

Apache HBase is a database that is designed to scale up to thousands of machines and manage billions of rows. It is successfully being used by web companies, banks, medical companies, only to name a few.

The first part of this presentation will be about HBase’s basics. We will start with the data model, which is quite different from the more common relational model, but very close to Google’s Bigtable with tables and families of columns. Then we will investigate the architecture and see what makes it so scalable at both the file system layer with Hadoop and at the database level with HBase.

The second part of the talk will be about use cases. The presenter will showcase the work done at his previous company where he was managing multiple HBase clusters of over a hundred nodes that were used for live web serving and batch processing. The challenges faced included not only doubling the hardware to fit the growing data set but also to accommodate a 5x increase in queries. Use cases from other companies that the presenter interacted with will also be presented in the interest of exploring different workloads and environments.