↓Skip to main content
  1. System Design Components/

System Design Core Index #

This is the entrypoint note.

Do not start by browsing every cheat sheet. Start here and only open the next note the current question needs.

The Core Stack #

Use this order:

  1. What kind of system is this?
  2. What is on the hot path vs cold path?
  3. How does one important unit evolve over time?
  4. What invariant/failure/repair story matters?
  5. Only if needed: where does load or backlog concentrate?

Primary Notes #

These are the only notes that should be in your default working set.

1. Archetype Classifier #

infra-archetype-taxonomy-reference.md

Use for:

  • prompt classification
  • dominant correctness center
  • truth / transitions / ownership / repair

Open first when:

  • the prompt is unfamiliar
  • you need to decide what kind of system it is

2. Hot / Cold Path Lens #

hot-path-cold-path-design-note.md

Use for:

  • architecture presentation
  • what belongs on the hot path
  • what should be pushed to cold/background paths

Open second when:

  • you need the architecture story
  • you need to explain why a mechanism belongs on or off the request path

3. Process Trace Lens #

process-trace-comparison-sheet.md

Use for:

  • process-heavy systems
  • workflows
  • schedulers
  • queues
  • control loops
  • routing/admission

Open when:

  • the problem is mainly about progression, ownership, retries, timeouts, repair

4. Storage Trace Lens #

storage-trace-comparison-sheet.md

Use for:

  • storage-heavy systems
  • replication
  • durability
  • visibility
  • freshness
  • derived views

Open when:

  • the problem is mainly about truth, versions, replicas, commit, stale reads

Secondary Notes #

These are useful, but only after the primary stack has already simplified the problem.

Distribution / Performance #

distribution-view-cheat-sheet.md

Use only when:

  • skew
  • backlog concentration
  • hot shards
  • retry storms
  • saturation

Do not start here.

Invariant / Failure / Repair Drills #

invariant-failure-repair-drill-sheet.md

Use when:

  • practicing
  • turning a prompt into failure -> invariant -> mechanism -> repair

NFR / Capacity #

infra-archetype-nfr-practice-worksheet.md

Use when:

  • you need concrete sizing
  • you need capacity units
  • you need to turn QPS into shards/workers/replicas

Paper Trace Notes #

Use when:

  • you want concrete system anchors
  • you want to see the trace lens instantiated in real papers

These are examples, not starting points.

Simple Workflow #

If the prompt is unclear #

  1. open archetypes
  2. classify the system
  3. identify storage-first or process-first center

If the prompt is process-heavy #

  1. open hot/cold path note
  2. identify hot path
  3. open process trace sheet
  4. derive unit state, owner/view, monotonic marker, validity boundary

If the prompt is storage-heavy #

  1. open hot/cold path note
  2. identify hot reads and writes
  3. open storage trace sheet
  4. derive truth locus, replica/derived state, durability/freshness, visibility

If the problem is really about load/skew #

  1. do the normal core stack first
  2. only then open distribution view

What To Ignore By Default #

Do not try to keep every note active in working memory.

Ignore by default:

  • overlays
  • phrase sheets
  • paper notes
  • drill notes
  • specialized mappings

These are lookup material, not your primary framework.

Compression #

Your working mental stack should be only:

  • archetype
  • hot/cold path
  • process or storage trace
  • failure/invariant/repair
  • distribution only if needed

That is enough.

←→ →←