Teaching workspace

System Design

Taught through documented architectures from Facebook, Discord, Amazon, Figma, Notion and PostgreSQL — every claim traced to a primary source. One win per lesson, retrieval practice at the end, and earlier material mixed back in on purpose.

Lessons

1. 01 The Replica Time Machine Replication & staleness Facebook · Scaling Memcache (NSDI ’13)
2. 02 Choosing Your Failure Mode Partitioning & tail latency Discord · 2017 and 2023
3. 03 The Dangerous Moment Cache misses & leases Facebook · §3.2.1
4. 04 Retries Are Selfish Timeouts, retries, backoff, jitter AWS Builders’ Library
5. 05 The Dial Quorums & the availability trade Amazon · Dynamo (SOSP ’07)
6. 06 The Cheap Move First Sequencing the scaling moves Figma
7. 07 Five Minutes Migrating a live database Notion
8. 08 The Shape of an Index Storage engines · Module 1 PostgreSQL 18 documentation
9. 09 Writing Fast and Reading Slow Storage engines · Module 1
10. 10 The Three Amplifications Storage engines · Module 1
11. 11 What an Index Costs Storage engines · Module 1
12. 12 Rows or Columns Storage engines · Module 1
13. 13 The Compatibility You Owe Data on one machine · Module 1
14. 14 What ACID Promises Transactions · Module 2
15. 15 Snapshots Without Locks Transactions · Module 2
16. 16 The Anomalies Transactions · Module 2
17. 17 Serializable and Its Price Transactions & isolation · Module 2
18. 18 The Coordination Tax Distributed transactions · Module 2

The full curriculum plans 13 weeks and roughly 120 lessons — storage engines, transactions, consensus, async work, the network edge, reliability, observability, sixteen real architectures, and a final module on design docs and running a review. Every lesson names its primary source before it is written.

Reference sheets

The compressed essence of each lesson, built to be printed and re-read rather than worked through.

• The Cache Miss Playbook — Two failures live in the gap between a miss and the write-back. They look alike on a dashboard and need different fixes.
• The Migration Playbook — The same four phases whether you are sharding a database, changing a column type, or replacing a queue. Load phases 1–3 until phase 4 has almost nothing to do.
• The Partitioning Playbook — Choosing a partition key, and recognising the outage when it was the wrong one.
• The Quorum Playbook — Three numbers, one inequality, and a long list of things that inequality does not mean.
• The Retry Playbook — Timeouts create retries, retries multiply, and backoff without jitter just gives the spike a rhythm.
• The Scaling Ladder — Take the cheapest rung that clears your runway requirement. Every rung you climb permanently removes something you used to get for free.
• The Staleness Playbook — Replicas, lag, and the four ways to make a user see their own write. Print it; it is designed to survive a black-and-white printer.