Review of Contemporary Database Design and Implication for Big Data

Data Modelling

Data modelling is a design discipline with the task of analyzing the business requirements followed by the design according to the requirements analysis outcomes. Data modelling, the first step in designing a database, refers to the representation of the data required to support a process or a set of processes (Simsion & Witt, 2005). Data modelling is an essential part of the design process and the development of a data system. Data modelling provides techniques for describing the real-world information requirements in an understandable manner to the users as well as support designers to implement the information requirements into a physical database system. Data modelling is an iterative progressive process which starts with understanding the problem domain by collecting and analyzing details about data elements and their suitability for supporting the business processes. The next step is ensuring that the results of the requirements definition are fully implemented as data contained in the database. Based on the requirements analysis, a proper database system will be selected. The final data model is actually a blueprint that contains all the instructions to build a database that meets all end-user requirements (Rob & Coronel, 2009). The data model can be fitted at any time during the database design life cycle. Therefore, it can be produced before, after, or in parallel/blended to the process model. For instance, for process-driven approaches, the focus is mainly on the process model; the data modelling process starts by identifying all the processes and their required data. Then the data model is designed to support the specific data requirements of a particular process. However, for data-driven approaches, where the data model is developed before the detailed process model, which promotes the reusability of data, a consistent set of definitions is established for data and language to classify the data. In practice, it is impossible to develop a data model without investigating the processes or developing a process model without considering the data. Therefore, parallel/blended approaches are the ideal choice for dealing with the interdependency of data and process modelling (Simsion & Witt, 2005).