Towards Self-Organizing Cloud Polyglot Database Systems

The demand for database OPEX reduction emerging from strict and heterogeneous storage requirements of modern cloud applications motivates shifting the contemporary manual and semi-automated approaches into an autonomic database management framework. As a first step towards fulfilling that vision, we introduce the Self-organizing Polyglot Persistence Architecture encompassing the key elements required for modeling self-adaptive capabilities into distributed database environments. We then implement SPPA-derived infrastructure framework based on services provided by OpenStack private cloud platform. Our experimental studies comprising various database scaling scenarios in the context of diverse database technologies (MariaDB Galera Cluster, Cassandra) confirm that the proposed SPPA approach enables cloud operators with a modular, declarative and flexible end-to-end solution for introducing adaptive database behaviors in the cloud.