Introduction to System Design

System design is the process of defining the architecture, components and data flow of a system so that it meets its requirements at the required scale.

A repeatable approach​

1. Clarify requirements: functional (what it does) and non-functional (scale, latency, availability, consistency).
2. Estimate: users, requests per second, storage, bandwidth.
3. Define the API and the core data model.
4. Sketch the high-level design: clients, services, storage, caches, queues.
5. Deep dive into the bottlenecks and trade-offs.
6. Review: failures, monitoring, security, cost, evolution.

Non-functional requirements​

Availability in "nines"​

Back-of-the-envelope estimation​

Useful numbers:

daily_users = 10_000_000
requests_per_user = 20
qps = daily_users * requests_per_user / 86_400
peak_qps = qps * 3                      # typical peak factor
print(round(qps), round(peak_qps))      # ~2315, ~6944

Trade-offs to name explicitly​

• Consistency vs availability (CAP).
• Latency vs throughput.
• Read-optimised vs write-optimised designs.
• Simplicity vs flexibility; build vs buy.