Course contentsShow
← System Design
Lesson 319 of 1,9199. SQL vs NoSQL TradeoffsFree lesson

Decision Framework: Data Model First

Evaluate data structure, relationships, and query patterns as primary factors for choosing SQL or NoSQL.

Decision Framework: Data Model First

You'll learn: How to choose between SQL and NoSQL by analyzing your data's natural shape, relationships, and access patterns before anything else.


When choosing a database technology, many engineers jump straight to features or popularity. But the data model — how your data is structured and queried — should be your primary decision factor.

The Three-Question Framework

Start by examining your data through three lenses:

1. Data Structure Is your data highly structured with fixed schemas (user profiles, financial records), or flexible and schema-less (event logs, product catalogs with varying attributes)? SQL excels at structured data with strict types; NoSQL handles semi-structured or unstructured data naturally.

2. Relationships Do entities have complex, multi-level relationships that require frequent JOINs (orders → customers → addresses → payments)? SQL's relational model and JOIN operations make this elegant. If your data is mostly independent documents or requires minimal cross-entity queries, NoSQL's simpler access patterns fit better.

3. Query Patterns How will you read and write data? SQL shines with ad-hoc queries, complex aggregations, and analytical workloads. NoSQL optimizes for predictable access patterns (key-value lookups, simple queries) and write-heavy workloads where you denormalize data for specific read patterns.

The Decision Path

  • Strong relationships + complex queries + structured schema → SQL (ACID guarantees are a bonus)
  • Document/hierarchical data + predictable access + flexible schema → NoSQL (BASE properties support scale)
  • Hybrid needs → Consider using both (polyglot persistence)

Key Takeaway: Choose your database by matching your data's natural structure, relationship complexity, and query patterns to the strengths of SQL (relational, structured, flexible queries) or NoSQL (document-oriented, scalable, pattern-specific access) — not by hype or default.