The trade-offs
Replication and partitioning, caching, asynchronous processing and reliability: the four areas where designs are won or lost.
Lessons
- 01Replication, partitioning and consistencyCopies buy availability and read scale, partitions buy write scale, and both cost you some form of consistency.
- 02Caching and its failure modesCaches are the cheapest latency win and the most common source of subtle bugs: plan invalidation, stampedes and cold starts up front.
- 03Queues, streams and idempotencyAsynchronous work is at-least-once in practice, so every consumer must be idempotent and every pipeline needs a poison-message plan.
- 04Reliability: SLOs, timeouts, retries and regionsDecide how reliable the system must be, then spend that error budget deliberately with timeouts, retries, isolation and graceful degradation.