The correct answer is Assign time to live values to keys so they expire automatically.

Setting expirations on keys ensures that entries are removed automatically when their time to live elapses and memory is reclaimed without waiting for the cache to fill completely.

Redis supports key expiry and it removes expired keys lazily when they are accessed and it also runs an active expiration cycle in the background so expired keys are purged and memory is freed periodically rather than only under memory pressure.

Enable an eviction policy such as allkeys-lru for the cache is not the best answer because eviction policies are reactive and run only when the cache nears its memory limit. They do not proactively remove entries at a specified time to live.

Periodically flush the entire cache is incorrect because flushing removes all keys and causes widespread cache misses and it is a blunt instrument rather than a targeted, proactive expiry of stale entries.

Record last access timestamps and run a background job to prune idle entries is incorrect because it requires extra application logic and storage and it adds overhead. Redis already offers native TTL support so this custom pruning is usually unnecessary and more complex.

Full AWS Practitioner Certification Question