When you PUT data to your origin via Cloud State, it gets eventually replicated to your own data store (for example, your S3 bucket).

Until it gets to your data store, all the GETs to the value will be served using the intermediary cache in Cloud State, to ensure that your clients don't have non-repeatable reads.

After the value gets replicated, it is dropped from the intermediary cache, but will respect the cache.ttl setting of the origin.

For example, if there's no cache.ttl in the origin, and the value exits the replication process, Cloud State will not store that data any longer.