Enhancing Query Support in HBase via an extended Coprocessor Framework

Abstract

Data is growing at an unforeseen rate, with TBs being generated daily. A large part of this data is unstructured in nature. This has pushed the traditional techniques of storing it in relational databases to its limit and new alternatives are necessary. Cloud databases have emerged as a viable candidate and have been gaining popularity due to their high scalability and availability. However, as yet, they lag behind RDBM systems in terms of the support to developers for querying the data. The problem of developing frameworks to support flexibe data queries is a very active area of research. In this work we consider HBase, a popular cloud database, inspired by Google’s BigTable data structure. Relying on the Coprocessor feature of HBase, we have developed a framework that developers can use to implement aggregate functions like row count, max, min, etc. We further extended the existing Coprocessor framework to support Cursor functionality so that a client can incrementally consume the Coprocessor generated result. We demonstrate the effectiveness of our extension by comparatively evaluating it against the original Coprocessor framework with four queries on three different data sets.