In the time of big data and data products, proper data architecture is an absolute essential. Without a data architecture working as a backbone, data cannot be used to generate value for your business. But what is needed to build a data architecture? Besides defined workflows and an overarching governance, data architectures need to have a working technical foundation. This article provides insights on how data modelling builds the foundation for such a data architecture.
What is Data Modelling?
Data modelling is the practice of using words and symbols to represent data and how it flows in a simplified representation of a software system and the data pieces it includes. Data models can serve as a roadmap for creating new databases or reengineering an existing one. Furthermore, it is an essential aid for designing a new data architecture. Overall, data modelling assists an organization in successfully using its data to satisfy business information demands.[1]
Data modelling is an important aspect of data management. It helps to detect information needs for distinct business processes by giving a visual representation of data sets and their business context. The data pieces that will be included in applications as well as the database or file system structures needed to process, store, and manage the data are then specified.
A data model can be a flowchart that depicts data items, their properties, and the relationships that exist between them. Before any code is developed or architecture is designed, it allows data management and analytics teams to document data needs for applications and uncover mistakes in development plans.
Data Modelling vs. Architecture
Data modelling and data architecture can be seen as two key factors of data management. Both differ from each other, but their relationship has an enormous potential in data management.
Why can data modelling help to design and build a data architecture? Data modelling provides a micro view of data while data architecture has a macro focus on data.
Data modelling addresses how data is represented. It is concerned with data consistency and focuses on data correctness. Moreover, a data model is an effort to describe reality in a digestible manner. It is a representation of a small number of entities and their relationships. The goal is to visualize business concepts, relationships as well as the value of each entity.
Conversely, a company’s data architecture is providing a conceptual framework for overall data management and has a logical structure. It covers tools and platforms that will be used to store and analyse data. Data architecture looks at the data security and the infrastructure that houses that data. The whole organization’s data infrastructure is covered by data architecture.
Finally, data modelling and data architecture complement each other – a well-defined data model forms the basis for data architecture design.
Data architecture is a system and has a logical structure, therefore the logical components must be included for building and designing it. Moreover, the entire organization data infrastructure is covered by data architecture, so a representation of business and architecture must be provided.
Data Modelling Categories
The subject area of data modelling can be defined very broadly. If this broad area is divided into smaller sub-areas, the notations to be used can also be subdivided accordingly in order to better meet the respective requirements.
A data model is an abstract representation of the real-world things that interact in the business environment of an organization. It presents data entities, their properties, and the relationships between them. Data models can represent different levels of detail. Therefore, they can be divided into four categories/process levels: holistic, conceptual, logical, and physical.[2]
Holistic data models can provide an overview of an entire IT architecture or IT landscape that stands above the data itself. The holistic overview gives the opportunity to understand the data usage on an enterprise-wide level and to derive measures for strategic adaptions or data governance.
At a deeper level of detail, conceptual data models represent how data is used across business streams on an overarching perspective, by providing a technical overview of either the business or the architectural perspective. Nevertheless, conceptual data models are not bound to any technical restriction such as a limitation to a database or an application.
This detail comes into play when defining logical data models. These are required when defining relationships between two data entities and when creating data structures. These definitions are necessary for the implementation of the model, which is represented in a physical data model.
For the implementation and realization of data architecture the mostly used data model are either conceptual or logical. Within the boundaries of these two models, an entire data architecture can be defined.
At CAMELOT, we use data modelling as a foundation for building your data architecture. Thereby, the advantages of conceptual and logical data models are used as semantic data notations for the construction and design. But how to proceed from a working foundation? Read here about effective data architectures – challenges and solutions.
[1] What Is Data Modeling? – Definition from SearchDataManagement (techtarget.com)
[2] https://www.talend.com/resources/data-model-design-best-practices-part-2/
We would like to thank Geron Keller for his valuable contribution to this article.
