Source: diagnostic/diagnostic-quiz.md

Diagnostic Quiz — Baseline Assessment

Purpose: Gauge your current level so I can calibrate lesson depth and skip topics you already own. Time: ~15 minutes. Answer honestly — guessing hurts calibration. How to answer: Just type the question number and your answer in chat (e.g. Q1: b, Q2: True, Q6: I'd use X because...).

Format: 3 MCQ · 2 True/False · 3 Short answer · 1 Scenario · 1 Mini design.

Q1 (MCQ) — Load Balancer

A Layer 7 (L7) load balancer differs from a Layer 4 (L4) load balancer primarily because:

  • a) L7 is faster because it operates at a lower network layer
  • b) L7 can make routing decisions based on the content of the request (e.g. URL path, headers, cookies)
  • c) L7 is used only for TCP traffic while L4 is used for HTTP
  • d) L7 does not require SSL termination

Q2 (MCQ) — Caching

Which of the following is the strongest reason to add a cache in front of a database?

  • a) To make the database durable
  • b) To reduce read latency and offload repeated reads from the DB
  • c) To replace the database entirely
  • d) To make writes atomic

Q3 (MCQ) — Databases

Which situation most clearly favors a NoSQL database over a traditional relational (SQL) database?

  • a) You need multi-row ACID transactions across a financial ledger
  • b) You have highly structured data with complex joins
  • c) You need to store massive volumes of semi-structured data (e.g. user activity logs) with flexible schema and horizontal scale
  • d) You need strict foreign key constraints across many tables

Q4 (True/False) — CAP theorem

"In the presence of a network partition, a distributed system must choose between consistency and availability — it cannot have both."

Answer True or False, and in one line say why.

Q5 (True/False) — HTTP

"HTTP is a stateful protocol — the server automatically remembers information about the client between requests."

Answer True or False, and briefly justify.

Q6 (Short answer) — Latency

Roughly rank these operations from fastest to slowest (order matters, not exact numbers):

  1. Read 1 MB sequentially from SSD
  2. Round trip within the same datacenter
  3. L1 cache reference
  4. Read 1 MB sequentially from RAM
  5. Round trip California → Netherlands (over the internet)

Q7 (Short answer) — DNS

In 2–3 sentences, explain what DNS does and roughly what happens when you type www.google.com in your browser (up to the point the browser has an IP address).

Q8 (Short answer) — Scaling

What's the difference between vertical scaling and horizontal scaling? Give one advantage and one disadvantage of each.

Q9 (Scenario) — Choosing a data store

You're building a URL shortener (like bit.ly). You expect: - 100M new short URLs / month - 10:1 read-to-write ratio - Reads must be < 50 ms globally

Would you pick a SQL database (e.g. Postgres) or a NoSQL key-value store (e.g. DynamoDB / Redis-backed)? Justify in 3–5 lines. There's no single "right" answer — I'm grading your reasoning.

Q10 (Mini design) — Rate limiting

A public API needs to enforce: "each API key can make at most 100 requests per minute."

In 5–8 lines, sketch how you would implement this. Mention: - Where the counter lives - What happens when the limit is exceeded - One potential problem with your approach

When you're done

Paste your answers in chat like:

Q1: b
Q2: b
Q3: c
Q4: True — during a partition, you can't guarantee both...
Q5: False — HTTP is stateless...
Q6: 3 > 4 > 1 > 2 > 5
Q7: DNS translates...
Q8: Vertical = bigger machine...
Q9: I'd pick NoSQL because...
Q10: I'd use Redis with INCR + EXPIRE...

I'll review each answer, mark it, explain any misconceptions, assess your level, and then kick off Day 1 with the appropriate depth.