Data systems at scale
Reason about caches, replication, partitioning, streams, and coordination.
Add capacity without losing the meaning of the data
Two hundred readers miss the same cache entry. A replica has yesterday’s version. A single tenant overwhelms one partition while the fleet average remains low. Each case requires a different boundary and a different measurement.
Start with the cache and consistency models, then study projections, partition skew, and the larger design briefs. State coordination scope, freshness, and write authority. Local examples demonstrate protocols, while AWS adapters and real capacity measurements remain explicit deployment work.
Parts group related chapters. Each lesson has a chapter.lesson address, such as 4.07. Open a title below, or use Next to follow the reading sequence. Within a lesson, On this page lists its sections.
- 10.01
Protect shared storage with bounded cache loads and consistent reads
Concepts and examples
- 10.02
Trace LRU eviction before implementing its linked order
Concepts and examples
- 10.03
Implement LRU without an ordered-map helper
Concepts and examples
- 10.04
Expiring key-value store
Concepts and examples
- 10.05
Bound cache misses across instances and preserve fresh reads
Concepts and examples
- 10.06
Enforce revocation even when a CDN already has the content
Concepts and examples
- 10.07
Count event-time windows with late arrivals
Concepts and examples
- 10.08
Keep job status current without repeating completed work
Concepts and examples
- 10.09
Distribute a hot tenant while preserving event identity and ordering
Concepts and examples
- 10.10
Synchronize files with resumable uploads and conflicts
Concepts and examples
- 10.11
Search documents without leaking revoked content
Concepts and examples
- 10.12
Compute trending topics from duplicate and late events
Concepts and examples
- 10.13
Ingest events with durable acceptance and replay
Concepts and examples
- 10.14
Build a durable crawler with per-host limits
Concepts and examples
- 10.15
Build typeahead with stale-response protection
Concepts and examples
- 10.16
Protect a database with versioned cache fills
Concepts and examples
- 10.17
Implement replicated writes and fenced leadership
Concepts and examples
- 10.18
Aggregate click events with late arrivals and reconciliation
Concepts and examples
- 10.19
- 10.20
Stage 3: Add durable jobs and bounded caching
Projects and practical assessment