From relations to multi-dimensional maps: a SQL-to-HBase transformation methodology
Diego Serrano, Eleni Stroulia · Computer Science and Software Engineering · 2016
In this paper, we describe a methodology for migrating applications relying on relational databases to HBase backends. Our methodology includes (a) a SQL-to-HBASE data-schema migration step, and (b) a transformation of the application SQL queries to equivalent sequences of HBase API calls. Our data-schema migration method relies on a set of HBase-organization guidelines to drive a four-step data-schema transformation process. Some of these guidelines are query-agnostic: we defined them based on related literature regarding the desired properties of the HBase organization. Other guidelines are query-aware: we formulated them to incorporate data-access paths, extracted from query logs, in order to improve the quality of the transformation and the eventual access efficiency of the HBase repository. Our transformation method maintains a mapping between source and target schema that is used to create sequences of HBase API calls, equivalent to SQL queries in the relational database. We illustrate and validate our method with a case study and a comprehensive performance evaluation.