The landscape of big data is defined by an overwhelming array of choices. When architecting a new system, one of the most critical yet complex challenges engineers face is selecting the right data model. Should the system rely on a traditional SQL relational database, or is a NoSQL solution—like a document store or a column-oriented database—more appropriate?
Historically, this decision has often been driven by developer familiarity, industry trends, or "gut feeling" rather than quantifiable metrics. However, as data systems grow exponentially, intuition is no longer a scalable or responsible strategy. My current research focuses on bridging this gap, exploring methodologies that bring systematic, automated, and sustainable decision-making to data model selection.
The Shift Toward Automated Modeling
A cornerstone of this research area is the exploration of frameworks like DaMoOp (Data Model Optimization), recently published in Information Systems (2025) by Jihane MALI et al. Frameworks of this nature tackle the architecture problem head-on by eliminating the guesswork.
Instead of manually evaluating disparate database technologies, the approach begins at the highest level of abstraction: the conceptual model. From there, the system automatically generates relevant logical data models tailored to the specific use case. This automation is a massive leap forward, but the true innovation lies in how these generated models are evaluated.
A Multidimensional Approach to Optimization
Choosing a data model is an exercise in trade-offs. To make an objective decision, we must quantify those trade-offs. The research advocates for a multidimensional cost model that evaluates generated architectures across three distinct axes:
- Time Cost: Evaluating query latency, throughput efficiency, and overall system performance.
- Financial Cost: Calculating the operational overhead, including storage, compute resources, and maintenance.
- Environmental Cost: Measuring the energy consumption and carbon footprint of the data architecture.
Engineering for Sustainability
The inclusion of environmental cost represents a vital paradigm shift. We operate in an era where data centers consume an enormous and rapidly growing share of the world's energy. Traditionally, "performance at all costs" was the guiding principle of data engineering. Today, baking sustainability directly into the data modeling decision process is no longer optional; it is a necessity.
By integrating frameworks like DaMoOp into the architectural planning phase, we can transition toward responsible data engineering. This research underscores a future where database design is not just about building faster or cheaper systems, but building systems that actively account for their real-world environmental impact.
As the industry continues to evolve, the intersection of database research and sustainable innovation will be critical.