SQL-to-NoSQL Schema Denormalization and Migration: A Study on Content Management Systems

Content management systems (CMSs) are able to let people, who have no technical skill to manage websites, rapidly create, edit, organize and publish online contents. For example, CMSs are often used to establish e-commerce online shops, community portals, personal blogs, or organization websites. Thus, CMSs play the role of collecting data and then can be easily extended to become intelligent Internet systems. Most popular CMSs, e.g., Word Press and Joomla, rely on relational databases or SQL databases, e.g., MySQL, PostgreSQL, etc. However, due to the explosive growth of huge data, the well-structured characteristic of SQL databases may limit the scalability to handle horizontal scaling. Therefore, regarding the flexibility and the feasibility of parallel processing, cloud computing takes Not Only SQL (NoSQL) databases into consideration. How to migrate the original CMS software from SQL databases to NoSQL databases becomes one emerging and critical research issue. This paper is motivated to propose an autonomous SQL-to-NoSQL schema denormalization and migration. At the same time, this paper also evaluates the proposed mechanism on the realistic CMS software. Based on our experimental results, our mechanism not only helps the current CMS software migrate into the cloud platform without re-designing their database schemas, but also improves at least 45% access performance after schema migration.