Tempo by Grafana
You'll learn how Tempo uses object storage to provide cost-effective distributed tracing that seamlessly integrates with Grafana.
What Makes Tempo Different
Tempo is Grafana's distributed tracing backend designed with a radical simplicity in mind: it stores traces directly in object storage (like S3, GCS, or Azure Blob) rather than relying on expensive indexing systems. Think of it like storing entire books (traces) in a warehouse without creating a card catalog—you find them using metadata from your logs and metrics instead.
Cost-Effective Architecture
Unlike traditional tracing backends that index every span attribute (which becomes costly at scale), Tempo only indexes trace IDs. The complete trace data lives in cheap object storage as compressed blocks. This approach reduces storage costs by 10x or more compared to fully-indexed solutions.
When you need a trace, Tempo retrieves it directly using the trace ID—typically found in your logs via correlation IDs or discovered through your metrics.
Integration with Grafana
Tempo shines brightest within the Grafana ecosystem:
- Unified Interface: Query traces alongside logs (Loki) and metrics (Prometheus) in a single pane
- TraceQL: Use Tempo's query language to search traces by service, duration, or status
- Automatic Linking: Click from a log line with a trace ID straight into the full trace visualization
- Service Graph: Generates dependency maps automatically from trace data
The architecture is elegantly simple: your applications send traces via OpenTelemetry, Tempo writes them to object storage in chunks, and Grafana pulls everything together for correlation and analysis.
Key Takeaway: Tempo sacrifices deep indexing for massive cost savings by storing traces in object storage and relying on trace ID lookups, making it ideal for high-volume tracing when integrated with Grafana's observability stack.