Data freshness is the degree to which data reflects the latest available state of its underlying source. Caching reduces access latency by reusing stored results, but the cached copy can lag behind the backing store. section 6.2
Freshness becomes a design trade-off when cached data is retained:
- Non-transactional caches may expose stale data because they provide no isolation guarantees. section 6.2
- Persistent caches survive restarts and avoid cold starts, but can increase consistency complexity between the cache and its underlying store. section 6.2
- Time-to-live (TTL) bounds how long an entry remains valid: values that are too low reduce hit ratio, while values that are too high allow stale reads. section 6.11
Cached REST API result
A client can serve a previously computed REST API result from memory instead of making another network request. This lowers latency, but the result may no longer match the latest state in the backing store. section 6.2
