CacheInvalidation Strategies
The hard part of caching: TTL-based expiry, write-through, cache-aside, event-driven invalidation.
Why we need this / what value this brings
A cache that serves stale data indefinitely is often worse than no cache at all — invalidation is what makes caching safe to use.
When to use this
Design invalidation at the same time you add the cache, not as an afterthought once a stale-data bug is reported.
How to use or implement this
Prefer TTL-based expiry for simplicity; use explicit invalidation (on write) only where staleness genuinely can't be tolerated.
Research questions
- 'There are only two hard things in CS...' — what's Localz's actual invalidation story once a provider updates a listing?
Empty folder — drop notes, links, and findings here as you research.