2.2. Workload Types: The Intent of Data

💡 First Principle: Every database is built with a fundamental question in mind: "What will you ask of this data?" Systems optimized for writing individual records safely (transactional) use completely different architectures than systems optimized for reading massive aggregates (analytical). It's like the difference between a bank vault and a library—both store valuable things, but their designs reflect opposite priorities. You cannot optimize for both simultaneously with a single engine. Attempting to do so creates a system that does neither well.

Consider what happens when you violate this principle: Run a complex analytical query scanning 50 million rows on your production OLTP database, and you'll slow down every customer transaction. Your checkout pages freeze, shopping carts time out, and customers abandon purchases. That single report query could cost thousands in lost sales.

Scenario: A bank has two needs: (1) Process individual ATM withdrawals with guaranteed accuracy—the customer's balance must never be wrong. (2) Generate monthly reports showing withdrawal trends across all 10 million customers. These two needs require fundamentally different database architectures.

Understanding workload types helps you select the right Azure services for each use case.