HLD Cheat Sheet
Intermediate · 15 min read · High-Level Design, Revision
Numbers, trade-offs and decision tables on one page for last-minute revision.
A one-page revision sheet for the night before (or the hour before) a design interview. Every entry links back to the lesson that explains it in depth.
The interview flow
Decision template: requirement → options → choice → cost. (Framework)
Numbers
(Estimation)
Choosing storage
(SQL vs NoSQL, Blob Storage, Batch & Stream)
Scaling data
Consistency cheat sheet
(CAP, Consistency, Transactions)
Communication
Reliability toolkit
(Fault Tolerance, Rate Limiting, Consensus, Observability)
Case study one-liners
Final-minute checklist
Requirements confirmed · numbers drive decisions · every functional requirement has a path through the diagram · consistency chosen per data type · no single point of failure · retries are idempotent · hot keys handled · failure walk-through done · observability and security mentioned · trade-offs stated out loud.
Practice
Write your answer before opening the suggested one. If you mark it "Partly" or "Missed it", it comes back in your review deck.
Key takeaways
- Run every design in the same order: requirements → estimates → API → data model → high-level design → deep dives → wrap-up.
- Let numbers drive decisions: a day is about 10^5 seconds, 1M requests a day is about 12 QPS, and you always size for peak.
- Choose storage and consistency per data type; money and locks lean strongly consistent, feeds and counters can be eventual.
- Scale reads with caches and replicas, writes with sharding and queues; remove single points of failure and make retries idempotent.
- Finish with failures, hot keys, observability, security and the trade-offs you made, said out loud.