Learning Maps, a Napkin Films series

A Napkin Films series

Learning Maps

System design taught as story. One concept per episode, each a room on a map.

Learning Maps turns the hard concepts behind distributed systems, and the AWS Solutions Architect (SAA-C03) exam, into short story-grade films. Each episode is one room on a map: a single idea introduced through two librarians and their branch libraries, a forced choice, the named concept, and a retrieval beat where the question hangs in real silence before the answer lands.

Every lesson splits the durable principle (the physics that never changes) from the rented label (this year’s AWS marketing name). You learn the principle as the spine and hang the buzzword on it as a sticker you can peel off when the vendor rebrands. The certs are the scaffolding; the principle is the building that stays standing after the exam is forgotten.

It started as my own SAA-C03 study, but the spark was O’Reilly’s System Design on AWS. It made me want to learn the systems underneath the exam, not just the service names, and to carry that thinking straight into my day work in platform engineering and systems-analysis mapping. The principle layer is the same whether you are passing a cert, designing a platform, or charting a system you already run, which is the whole point of making it legible.

The films are audio-first, built to survive partial attention: they work on a walk with your eyes closed, and the hand-drawn napkin map rewards watching without ever requiring it.

6 of 25 episodes released

All Napkin Films →

The Data Wing

Foundations to consensus: how data stays correct when it lives in more than one place.

  1. Episode 1 The Forced Choice CAP and PACELC: the consistency-versus-availability tradeoff you cannot escape. Watch · 9:05
  2. Episode 2 The Dial Quorum and R + W > N: tuning read/write voting instead of flipping a switch. Watch · 6:22
  3. Episode 3 Just Enough Agreement Read-your-writes and monotonic reads: the consistency you actually need. Watch · 6:16
  4. Episode 4 The Cache Cache, hit, miss, TTL, and the hard part: invalidation. Watch · 5:32
  5. Episode 5 The Branch Library Replication: leader/follower, sync versus async, Multi-AZ versus read replicas. Watch · 7:23
  6. Episode 6 The Shard Partitioning: shard key, the hot partition, range versus hash. Watch · 7:27

The Systems-Talking Wing

Coordination to the edge: how independent services agree, decouple, and stay up.

  1. Episode 7 The Vote Consensus and leader election: split brain, majority, terms, Paxos and Raft. In production
  2. Episode 8 The Ring Consistent hashing: the hash ring, virtual nodes, the mod-N stampede. In production
  3. Episode 9 The Book Drop Queues and decoupling: SQS, SNS pub/sub, idempotency, dead-letter queues. In production
  4. Episode 10 The Greeter Load balancing: ELB/ALB/NLB, health checks, statelessness, horizontal scale. In production
  5. Episode 11 The Breaker Resilience: timeout, retry with backoff, circuit breaker, bulkhead, fallback. In production
  6. Episode 12 The Allowance Rate limiting: token bucket, 429 with Retry-After, per-caller versus global. In production
  7. Episode 13 The Outpost CDN and edge: edge locations, origin, hit/miss, CloudFront, latency routing. In production

The Security Wing

Permissions to encryption: making the blast radius small on purpose.

  1. Episode 14 The Keyring Least privilege and IAM: deny by default, scoped roles, temporary keys. In production
  2. Episode 15 The Gauntlet Defense in depth: layered independent controls (SG, NACL, WAF, IAM, encryption). In production
  3. Episode 16 The Lobby Trust boundaries and attack surface: public versus private subnets, VPC, NAT, bastion. In production
  4. Episode 17 The Cipher Encryption at rest and in transit: TLS, KMS, key management. In production

The Cost Wing

Elasticity to architecture: why the bill is a design decision.

  1. Episode 18 The Ballroom Pay for what you use: on-demand, auto scaling, serverless elasticity. In production
  2. Episode 19 The Promise Commitment pricing: on-demand versus reserved/savings versus spot. In production
  3. Episode 20 The Shelf Right-sizing and storage tiers: S3 Standard/IA/Glacier, Intelligent-Tiering. In production
  4. Episode 21 The Blueprint The bill is in the architecture: managed/serverless, TCO, data gravity and egress. In production

The Capstone Movement

Real systems, end to end, naming each principle as it gets used.

  1. Capstone 1 Designing a URL Shortener Read-heavy key-value lookup: DynamoDB plus cache plus edge; a distributed base-62 counter. In production
  2. Capstone 2 Designing a Social Newsfeed Fan-out on write versus read, and the hybrid matched to the shape of the account. In production
  3. Capstone 3 Designing a Game Leaderboard The right data structure deletes the query: a sorted set instead of a 50M-row scan. In production
  4. Capstone 4 Designing a Chat Application Real-time: WebSocket state contained in a thin gateway, registry, pub/sub, heartbeat-TTL. In production

Where the map goes next

The principles arc (the first four wings) is written end to end. The Capstone Movement then designs real systems, a URL shortener, a social newsfeed, a game leaderboard, a chat application, naming each principle as it gets used, so the whole codex reviews itself through practice.

Beyond that: a video pipeline (the opposite of chat, work too heavy to ever make you wait), a hotel reservation, a web crawler, and a stock-exchange finale, plus a standalone on observability.

Go deeper: AWS Is Math, Kubernetes Is Physics. The frame underneath the whole map: two ways of reasoning about infrastructure.