From Relational Database to Column-Oriented NoSQL Database: Migration Process

Wherever there is a need for the availability of services with a high traffic, maintaining at the same time a high performance, the blocking point often is the database. When one talks about high traffic systems or services and databases, one rarely hears of relational databases. Indeed, to ensure data consistency is expensive in time and is often inconsistent with performance. Since the relational model does not seem suitable in environments requiring large architectures and the ACID properties of bases generally do not allow to scale, a new movement was born from the initiative of cloud’s architects and other communities like Amazon and LinkedIn: NoSQL (aka: Not Only SQL). NoSQL databases are radically changing the architecture of the database that one used to see and thus allow to increase the performance and availability of services. Thereby, it seems useful to put forward a migration approach of conventional database to a NoSQL database. To do this, some limitations of relational RDBMS to dealing with large volumes of data is presented. Then the NoSQL technology and its strengths, issues and challenges are discussed. Finally, a migration from a relational database to column oriented NoSQL database was proposed as the aim of this paper. Keywords—NoSQL, Migration, column oriented database, data conversion, schema translation, column family, super column.