golang-samber-hot

To implement stale-while-revalidate and negative caching in a Go in-memory cache, use samber/hot with WithRevalidation and WithMissingCache configurations. This keeps data fresh while caching missing keys to prevent repeated backend lookups.

The best eviction algorithm for high-concurrency Go caching depends on access patterns; samber/hot supports LRU, LFU, TinyLFU, W-TinyLFU, S3FIFO, ARC, and SIEVE. W-TinyLFU often excels for mixed scan-heavy workloads.

To prevent cache stampede and deduplicate concurrent lookups in a Go application, configure samber/hot WithLoaders with built-in singleflight semantics. This collapses parallel identical requests into a single backend fetch.

To safely handle mutable values and sharded access in an in-memory Go cache, enable WithSharding for high concurrency and WithCopyOnRead to prevent race conditions. This ensures data isolation across concurrent read operations.

Yes, samber/hot supports Prometheus metrics for cache observability and WithJanitor for background TTL cleanup. Configuring WithTTL with the janitor ensures automatic eviction of expired keys without blocking read operations.

You should avoid in-memory caching in Go for extremely high-cardinality datasets that exceed available RAM or when strict cross-instance consistency is required, as samber/hot is designed for local, medium-to-low cardinality lookups to reduce backend load.