Management of Flexible Schema Data in RDBMSs - Opportunities and Limitations for NoSQL -
Zhen Hua Liu, Dieter Gawlick · Conference on Innovative Data Systems Research · 2015
1. ABSTRACT RDBMSs are designed to manage well-structured data requiring users to design a schema before storing and querying data. This is the ‘schema first, data later’ approach. However, there are significant amount of unstructured data and semi-structured data that cannot be effectively modelled this way. Even if certain parts of the data can be modelled using schema, the inclusion of all fields would typically lead to a very large schema with many optional fields and with frequent schema evolution as data instances vary widely and evolve fast. Obviously, these data requires the ‘data first, schema later/never’ approach. We call these data Flexible Schema Data (FSD) . In this paper, we describe the engineering principles and practices to manage FSD in RDBMSs to meet FSD’s unique requirements and challenges. We describe the limitations and issues of current practices and potential research opportunities. Having a single data platform for managing both well-structured data and FSD is beneficial to users; this approach reduces significantly integration, migration, development, maintenance, and operational issues.