What should I use for OpenTelemetry trace storage

For OpenTelemetry, the “right” trace storage depends mostly on your scale, retention needs, and whether you want a managed or self-hosted setup. OpenTelemetry itself doesn’t store traces; it exports them to a backend.

Good options

1. Grafana Tempo

Best if you want a cost-effective, scalable tracing backend.

• Designed specifically for traces
• Cheap to run compared to full indexing systems
• Integrates well with Grafana, Prometheus, Loki
• Good choice for Kubernetes / cloud-native environments

Use this if: you want long-term trace storage at scale and already use Grafana.

---

2. Jaeger

Best for simpler deployments and classic distributed tracing workflows.

• Very common and widely supported
• Good UI and tracing features
• Can use different storage backends
• Easier for smaller teams to adopt

Use this if: you want something familiar and straightforward, especially for smaller or medium setups.

---

3. Elastic Observability

Best if you already use Elasticsearch.

• Powerful search and correlation
• Unified logs/metrics/traces
• Good if you need advanced querying across observability data

Use this if: your org is already standardized on Elastic.

---

4. Commercial managed backends

Examples:

• Honeycomb
• Datadog
• New Relic
• Lightstep
• Sentry (for some trace use cases)

Use this if: you want low operational overhead and strong UI/analytics.

---

5. Cloud-provider tracing

Examples:

• AWS X-Ray
• Google Cloud Trace
• Azure Monitor / Application Insights

Use this if: you’re heavily committed to a single cloud and want native integration.

---

My practical recommendation

• Small/simple setup: Jaeger
• Best open-source scalable choice: Grafana Tempo
• Already on Elasticsearch: Elastic
• Want no ops: a managed SaaS backend

Important note

If you’re only asking about temporary storage or local dev, you can also use the OpenTelemetry Collector with an exporter to stdout/logs, but that’s not suitable for real trace retention.

If you want, I can recommend a backend based on your environment:

• Kubernetes or VMs?
• Cloud provider?
• Expected trace volume?
• Open source vs managed?