One concept at a time, with a constrained palette and validation that explains itself.
0 / 40 chapters
0 / 10 sections
Orientation - the learner needs to know what the discipline is, what the product expects of them, and how interview and production registers differ, before any real content.
The defining structural bet of the curriculum: the design method is taught on tiny, familiar systems using only the three primitive components, so the method itself is the only new material. Condensed from 11 chapters to 4 (3 mandatory + 1 optional) in Release 6.1.0-alpha Phase 10 - see .claude/docs/pending-6.1.0-poa.md.
One request is traced from keystroke to response so every Part 3 block later lands at a known address.
Just enough networking, naming, and routing to reason about everything above it - the trust perimeter and the request's first hops.
Why statelessness is the property that makes compute cheap to scale, and what it takes to actually scale it out.
What a database actually guarantees, when to reach for something other than SQL, and how copies and partitions change the guarantees.
The highest-leverage scaling tool (caching) plus the two derived-data problems it creates: caching at the edge and a second store that must stay in sync.
Everything so far has been synchronous request/response - work that happens later, and what happens when it fails.
Blobs, files, and what storage means once it spans machines - consistency models, quorums, and CAP stated honestly.
Object Storage
intermediate
File Storage
intermediate
Distributed Storage Concepts
intermediate
Timeouts, retries, rate limits, observability, and failover - the patterns that keep an already-correct system running.
Reliability Patterns
advanced
Rate Limiting
advanced
Observability
advanced
Fault Tolerance
advanced
Checkpoint · Building a Complete Backend
advanced
Checkpoint · Open System Design
advanced