Functional vs Non-Functional Requirements

Before you design anything, you need to know what you're building and how well it needs to work. These two questions split neatly into two categories that drive every architectural decision you'll ever make.

Functional Requirements: What It Does

Functional requirements describe the features — the visible behaviors users interact with. They answer: "What can a user do with this system?"

Non-Functional Requirements: How Well It Does It

Non-functional requirements (NFRs) describe the quality attributes — the invisible properties that determine whether the system actually works in production. They answer: "Under what conditions must it perform? What happens when things go wrong?"

The Iceberg Metaphor

Users see functional requirements. Engineers wrestle with the massive non-functional iceberg beneath:

The Key Non-Functional Requirements

NFR	Definition	Example	Measured As
Performance	How fast the system responds	Page loads in <200ms	Latency (p50, p95, p99)
Scalability	Ability to handle growth	Support 10M → 100M users	QPS, concurrent connections
Availability	System is up when needed	99.99% uptime (52 min downtime/year)	Nines (99.9%, 99.99%)
Durability	Data isn't lost	No uploaded photo is ever lost	99.999999999% (11 nines)
Consistency	All users see the same data	After transfer, balance updates everywhere	Strong/eventual/causal
Security	Protected from threats	Data encrypted at rest and in transit	Compliance, pen test results

Real-World Example: Twitter

Identify the Requirements