Monolith Cache Invalidation with Per-Module Redis Namespaces
Design monolith cache invalidation around module boundaries: Redis key namespaces, cache-aside races, version counters, after-commit events and safe TTLs.
14 min read
Tag
Design monolith cache invalidation around module boundaries: Redis key namespaces, cache-aside races, version counters, after-commit events and safe TTLs.
14 min read
Learn how event-driven eviction works across microservice caches, why events are only hints, and how TTLs, versioning, and outbox/CDC prevent stale cache.
17 min read
Design event-driven invalidation for in-process caches as bounded staleness: versioned events, TTL backstops, fan-out, race fixes and metrics.
20 min read
Add Redis to a monolith without stale reads: cache-aside with delete-on-write, after-commit invalidation, versioned keys, TTL backstops and safe rollout.
15 min read
Make distributed cache invalidation reliable with an outbox or CDC, a durable event log, versioned idempotent handlers, and TTLs that bound staleness.
13 min read
Build a Spring Boot two level cache with Caffeine in front of Redis, with pub/sub cache invalidation, staleness bounds, stampede control and failure handling.
17 min read
Choose between an in-process cache, Redis, or a two-level cache for a scaled monolith based on staleness budget, invalidation strategy, and operational cost.
14 min read
Design a monolith caching strategy across request, in-memory and Redis tiers, with commit-driven cache invalidation that keeps staleness explicit and bounded.
18 min read
How to drive cache invalidation in microservices from CDC or an outbox, handle late, duplicate and out-of-order events, and bound stale cache data.
13 min read
How to give each module in a modular monolith its own Redis key namespace, publish invalidation rules as a contract, and invalidate safely after commit.
14 min read
How to choose between an in-process Caffeine cache and a shared cache for a scaled-out monolith, covering cost, invalidation, stampedes and what to measure.
13 min read
Treat cross-service staleness as a producer-owned contract: derive TTLs from budgets, version invalidation events, fix races and measure cache drift.
14 min read