Rollover
Rollover creates a new index when the current one reaches certain conditions (age, size, document count). Elastic.Ingest.Elasticsearch supports several rollover approaches.
| Approach | How it works | Serverless? |
|---|---|---|
| Manual alias | Application manages write/read aliases and swaps them after indexing | Yes |
| Rollover API | Calls POST {alias}/_rollover with conditions |
Yes |
| ILM managed | ILM policy triggers rollover automatically based on conditions | No |
| Data stream lifecycle | Automatic rollover and retention via data stream lifecycle | Yes |
- Manual alias: best for batch sync patterns where you control when indices rotate (for example, catalog data)
- Rollover API: best when you want explicit control over rollover timing with conditions
- ILM managed: best for self-managed clusters that need multi-phase lifecycle (hot/warm/cold/delete)
- Data stream lifecycle: best for serverless and new projects that only need retention-based lifecycle