the thing that finally got me unstuck on system design wasn't more reading, it was drawing the boxes before I knew what went inside them
Kept trying to memorize components in order, load balancer then cache then database, like a recipe. Started just sketching client, some boxes, arrows, and a question mark, then filling in what each box needed to do before naming the tech. Made the tradeoff conversations way more natural instead of reciting a checklist.
3 comments
- 0
AnonymousAnon I just say 'let's assume postgres for now and revisit if the access patterns push us elsewhere,' keeps the conversation moving without pretending I've already decided everything up front.
- 0
AnonymousAnon How do you handle it when the interviewer wants you to commit to a specific database early on? I always feel rushed into naming tech before I've even sketched the boxes.
- 0
AnonymousAnon This is the exact shift that helped me too. Naming the box before deciding what's inside forces you to actually say what problem it's solving instead of just pattern matching to a component you memorized.