Iterate Based on Constraints
What you'll learn: How to systematically evolve your design by identifying bottlenecks and scaling points at each iteration.
The Core Idea
You've started with a simple design. Now what? Don't randomly add components—let the constraints guide you. Look at your non-functional requirements (scalability, latency, availability) and ask: "Where will this design break first?"
Think of it like planning a road trip. You don't upgrade to a truck immediately. You check: Will my sedan hold all the luggage? Will it handle mountain roads? Do I need more fuel range? Each constraint tells you what to improve next.
How It Works
-
Identify the bottleneck: Review your QPS estimate, storage growth, or latency targets. Which component can't handle the load?
- Database overloaded at 10K QPS? That's your bottleneck.
- Single server can't provide 99.9% availability? There's your constraint.
-
Propose a targeted solution: Address only that constraint.
- Add a cache to reduce database reads.
- Add load balancers and replicas for availability.
-
Re-evaluate: Check if new bottlenecks emerged.
- Cache helps reads, but what about writes?
- Replicas help availability, but how do they stay consistent?
-
Repeat: Each iteration solves one major constraint, revealing the next challenge.
This systematic approach prevents over-engineering (adding features you don't need) and under-engineering (missing critical requirements).
Key Takeaway: Let your estimated numbers and prioritized requirements reveal bottlenecks one at a time—each constraint you address guides the next design iteration, building complexity only where needed.