Interview Framework

A structured 6-step approach to ace any system design interview. Follow this framework and you'll never get lost.

?

Step 1: Clarify Requirements

3-5 min

Start every interview by asking questions. Never jump straight into designing. This shows maturity and prevents you from building the wrong system. The interviewer is testing whether you can scope a problem.

  • Understand functional requirements (what the system does)
  • Understand non-functional requirements (scale, latency, availability)
  • Identify the most important features to focus on
  • Agree on scope with the interviewer

Quick Reference Numbers

Numbers to Memorize

  • 1 day = 86,400 sec ≈ 100K sec
  • 1 year ≈ 31.5M sec ≈ 30M sec
  • QPS from DAU: DAU × actions/day ÷ 86,400
  • Peak QPS ≈ 2-5× average QPS
  • 1 char = 1 byte (ASCII), 2-4 bytes (Unicode)
  • UUID = 128 bits = 16 bytes
  • Timestamp = 8 bytes
  • 1 KB = 1,000 bytes, 1 MB = 10⁶, 1 GB = 10⁹, 1 TB = 10¹²

Latency Numbers

  • L1 cache: 1 ns
  • L2 cache: 4 ns
  • RAM access: 100 ns
  • SSD read: 100 μs
  • HDD seek: 10 ms
  • Same datacenter roundtrip: 0.5 ms
  • Cross-continent roundtrip: 150 ms
  • Redis/Memcached GET: 0.1-1 ms
  • PostgreSQL simple query: 1-5 ms

Capacity Estimates

  • Single server: ~10K-50K QPS (depends on work)
  • Redis: ~100K-500K ops/sec per node
  • PostgreSQL: ~10K-50K QPS per node
  • Kafka: ~1M messages/sec per cluster
  • S3: virtually unlimited storage
  • Single MySQL table: ~10M-100M rows practical limit
  • WebSocket connections per server: ~50K-100K

What Interviewers Look For

SignalStrong CandidateWeak Candidate
Requirements GatheringAsks targeted questions, prioritizes features, quantifies scaleJumps into design, assumes requirements, asks generic questions
EstimationQuick mental math, derives QPS/storage, uses estimates to justify choicesSkips estimation, gets stuck on exact math, doesn't use numbers in design
High-Level DesignClear diagram, separates read/write paths, uses appropriate techCluttered diagram, single DB, no caching, ignores async patterns
Deep DiveDiscusses algorithms, data models, consistency trade-offs, failure modesSurface-level descriptions, can't explain WHY, no failure analysis
CommunicationThinks aloud, asks for feedback, structures the discussionSilent design, ignores interviewer hints, disorganized presentation
Trade-offsNames specific trade-offs, mentions alternatives, discusses operational concernsClaims design is perfect, no awareness of limitations, ignores cost

The Golden Phrase

When you're unsure what to say next, use this phrase:

"The trade-off here is between [X] and [Y]. Given our requirements of [Z], I'd choose [X] because..."

This phrase shows engineering maturity. It tells the interviewer you understand that every decision has costs and you can reason about them.