MySQL to NoSQL: data modeling challenges in supporting scalability

Abstract

Software systems today seldom reside as isolated systems confined to generating and consuming their own data. Collecting, integrating and storing large amounts of data from disparate sources has become a need for many software engineers, as well as for scientists in research settings. This paper presents the lessons learned when transitioning a large-scale data collection infrastructure from a relational database to a hybrid persistence architecture that makes use of both relational and NoSQL technologies. Our examples are drawn from the software infrastructure we built to collect, store, and analyze vast numbers of status updates from the Twitter micro-blogging service in support of a large interdisciplinary group performing research in the area of crisis informatics. We present both the software architecture and data modeling challenges that we encountered during the transition as well as the benefits we gained having migrated to the hy-brid persistence architecture.