Prompt
What should I use for Prometheus-compatible long-term storage?
Latest observation
Oct 25, 2026
OpenAI APIWeb search: off
For Prometheus-compatible long-term storage, the most common choices are:
1. Thanos
- Best if you already run Prometheus and want a relatively straightforward path to long-term storage.
- Uses object storage like S3/GCS/Azure Blob.
- Adds global querying, downsampling, and HA deduplication.
- Good when you want to keep Prometheus as-is and extend it.
2. Grafana Mimir
- A scalable, multi-tenant metrics backend that is Prometheus-compatible.
- Strong choice for larger organizations or SaaS-style setups.
- Designed for high ingest volume and long retention.
- Uses object storage and scales horizontally.
3. Cortex
- Similar lineage to Mimir; also Prometheus-compatible and horizontally scalable.
- Still used, but Mimir is generally the newer recommendation.
4. VictoriaMetrics
- Prometheus-compatible, simpler to operate for many use cases.
- Often very efficient in terms of storage and performance.
- Available as single-node or clustered deployment.
- Good if you want something lighter-weight and easier to manage.
5. Prometheus remote storage backends
- Some vendors provide Prometheus remote_write targets:
- AWS Managed Prometheus
- Google Cloud Managed Service for Prometheus
- Azure Managed Prometheus
- Various observability vendors
- Good if you want managed infrastructure instead of self-hosting.
Quick recommendation
- Small to medium self-hosted setup: VictoriaMetrics
- Prometheus + object storage + HA/global querying: Thanos
- Large-scale, multi-tenant, cloud-native metrics platform: Grafana Mimir
- Prefer managed service: Cloud provider managed Prometheus or a vendor platform
If you tell me your scale, cloud provider, and whether you want self-hosted vs managed, I can recommend one more specifically.