Food for Thought | Data Model-Driven Strategy

This is the second blog post of the blog series about Titan 2.0.

In the early days of data management, companies didn't appreciate the complexity and potential of data as they do today. Initially, data collection and processing was siloed, with each department or team managing data for specific immediate purposes. Each use case was approached as an independent project, with disparate systems and technology, and its own interpretation of data, resulting in inconsistent metrics and definitions. These discrepancies in the way companies interpret data can create risks for decision-making. A strategy to prevent falling for these pitfalls is the data model-driven approach, which we will discuss in this blog post.

Data Model-Driven Approach

The data model-driven approach revolves around the idea that the data model itself should be the cornerstone of the data architecture, rather than being shaped around specific use cases. This doesn't imply ignoring business needs but rather anticipates the evolving nature of those needs. By focusing on creating a robust, flexible, and inherently scalable model from the outset, organizations can ensure that their data platforms are prepared for future requirements, not just current demands.

A practical example of transitioning from a use-case driven approach to a data model-driven approach is visualized in below diagrams. The example is focussed on four use cases that can be relevant in the production department of companies. For more information about the use cases, chech our previous blog posts: OEE, traceability, and waste reduction. The first diagram represents a use-case driven approach and the second diagram a data model-driven approach. In the first example, companies can have difficulties with standardizing the interpretation of the data.

Take the cost price calculation as example. Do you take into account the raw materials, labor, depreciation of machines and storage costs? What about the variable energy costs or energy return of solar panels? The data model-driven approach solves these questions by ensuring one representation of each 'data product' in the data platform.

Implementing a Data Model-Driven Strategy

Adopting a data model-driven strategy requires a shift in mindset from merely addressing current needs to envisioning a data architecture that can easily adapt to change. It involves:

• Investing in Design: Spending more time upfront on the design of the data model, considering not just present requirements but potential future adaptations.
• Embracing Flexibility: Choosing methodologies and technologies that support flexibility, such as anchor modeling and delta file schema evolution.
• Focusing on Governance: Establishing strong data governance practices to manage the evolution of data products effectively.

Conclusion

The shift from a use case-driven to a data model-driven strategy in data platforms represents a fundamental change in how organizations approach data architecture. By prioritizing flexibility, scalability, and future-readiness, businesses can create data platforms that not only meet today's needs but are also equipped to handle the challenges and opportunities of tomorrow.

FAQ