2. Core Data Concepts & Workloads (25-30%)

This phase expands the SHAPE and INTENT questions from Phase 1 into the vocabulary the exam actually uses: data representation, file formats, workload types, and the roles that run data platforms.

Why this matters: at 25–30% this is the joint-largest domain, and it is load-bearing for the three that follow. Every service question later in the exam is really a shape-and-workload question wearing a product name. Misclassify the data here and you will pick a defensible-sounding but wrong service in Phases 3, 4 and 5.

💡 The mental model: two independent axes. Shape is what the data looks like — structured, semi-structured, unstructured. Workload is what you do with it — transactional (many small writes) or analytical (few large reads). Each Azure data service occupies one cell of that grid.

⚠️ The trap: the two axes are independent, and the exam exploits candidates who collapse them. "Structured" does not mean "transactional" — a data warehouse is highly structured and purely analytical. "Non-relational" does not mean "unstructured" — a JSON document is semi-structured and perfectly queryable. Read for the shape and the intent separately, because a question will often give you one of each and expect you to combine them.