Time-series data (logs, metrics, events) grows forever. Elasticsearch manages it with: Feature Role Index template settings + mappings applied automatically to new indices matching a pattern Component templates reusable building blocks for index templates Data stream one name for append-only data backed by hidden rolling indices ILM policy lifecycle phases: hot → warm → cold → frozen → delete Rollover start a new backing index at a size/age threshold Typical ILM policy hot: rollover at 50 GB per primary shard or 1 day warm (7 d): shrink, force-merge, move to cheaper nodes delete (30–90 d) Shard sizing rule of thumb: 10–50 GB per shard; too many small shards waste heap.