What Is It?
Tracing Service is a fully managed, OpenTelemetry-native distributed tracing solution that lets you ingest, store, and analyze traces generated by your applications. You follow a request across multiple services, find latency bottlenecks, and analyze failures in Grafana. Together with Monitoring Service and Logging Service, it completes the observability triad, so traces, logs, and metrics are correlated in a single Grafana interface. IONOS CLOUD provisions, scales, secures, and patches the tracing backend for you.
Quick Facts
| Aspect | Details |
|---|---|
| Type | Managed distributed tracing service (OpenTelemetry-native) |
| Backend | Grafana Tempo |
| Ingestion protocols | OTLP/HTTP, OTLP/gRPC |
| Authentication | One unique key per tracing pipeline, sent as apikey header (OTLP/HTTP) or apikey metadata (OTLP/gRPC) |
| Network requirement | Outbound HTTPS on port 443 |
| Retention | 30 days, fixed (cannot be customized) |
| Visualization | Grafana with an automatically provisioned Tempo data source |
| Query language | TraceQL |
| IAM privilege | Non-admin users need Access and manage Tracing to create, update, rotate keys for, or delete tracing pipelines |
| Uptime SLA | 99.5% |
| API | Tracing Service API v1 |
Ingestion Endpoints
| Protocol | Endpoint | Note |
|---|---|---|
| OTLP/HTTP | https://<tracing-ingestion-endpoint>/v1/traces |
Sends traces over HTTP |
| OTLP/gRPC | grpcs://<tracing-ingestion-endpoint>/v1/traces |
Sends traces over gRPC |
<tracing-ingestion-endpoint> is a unique hostname generated for each pipeline. It is not the management API host.
Regional API Endpoints (Pipeline Management)
| Location | Endpoint |
|---|---|
| Berlin | https://tracing.de-txl.ionos.com/pipelines |
| Frankfurt | https://tracing.de-fra.ionos.com/pipelines |
| London | https://tracing.gb-lhr.ionos.com/pipelines |
| Worcester | https://tracing.gb-bhx.ionos.com/pipelines |
| Paris | https://tracing.fr-par.ionos.com/pipelines |
| LogroƱo | https://tracing.es-vit.ionos.com/pipelines |
| Lenexa | https://tracing.us-mci.ionos.com/pipelines |
| Las Vegas | https://tracing.us-las.ionos.com/pipelines |
| Newark | https://tracing.us-ewr.ionos.com/pipelines |
Default Limits
| Limit | Default |
|---|---|
| Tracing pipelines per contract | 10 (contact IONOS CLOUD Support for a higher limit) |
| Maximum trace size (combined span data per trace ID) | 5 MB |
| Sustained ingestion rate | 15 MB/s |
| Burst allowance | 20 MB |
| Concurrent in-progress (live) traces | 10,000 |
| Rate-limit response | HTTP 429 (OTLP/HTTP), RESOURCE_EXHAUSTED (OTLP/gRPC) |
What You Can Do
Create Tracing Pipelines
A tracing pipeline is a regional instance of the service that provides authenticated ingestion endpoints. When you create one, you choose a name and one ingestion protocol (otlp-http or otlp-grpc). The service returns a unique pipeline ID, the pipeline state and timestamps, a trace ingestion key, a protocol-specific trace endpoint, and a grafanaEndpoint for trace exploration.
Send Traces from Any Environment
You can send traces from any environment if your exporter can reach the endpoint and uses the correct pipeline key. Supported source categories are application SDKs (OpenTelemetry SDKs for languages such as Go, Java, Python, JavaScript, and .NET), the OpenTelemetry Collector, service meshes and proxies that emit OpenTelemetry-compatible traces, and custom instrumentation that builds OTLP payloads. The service uses the open OTLP standard without proprietary agents.
Note: For components that emit Jaeger or Zipkin formats, place the OpenTelemetry Collector in front of Tracing Service as a translation layer.
Explore Traces in Grafana
The Tempo data source is provisioned automatically in your Grafana instance, so traces are captured without extra manual configuration once your exporter points at the pipeline endpoint. Use the Explore view to search and view traces, TraceQL for queries, and service-graph visualizations for dependency analysis.
Note: The Drill down view expects an existing trace context and can show an error if you start there.
Correlate Traces, Logs, and Metrics
In the same Grafana workspace, you can move from a latency span to the corresponding Loki logs (trace-to-log) or to related Mimir metrics (trace-to-metric) in the same time context.
Observe AI and LLM Workloads
Tracing Service supports the OpenTelemetry semantic conventions for generative AI (GenAI). It captures execution paths, tool calls, LLM (Large Language Model) latencies, and token consumption as span attributes alongside your infrastructure signals.
Rotate Pipeline Keys
Generate a new key for a pipeline through the API, for example when a key was accidentally shared, was lost, or a team member with access left the company.
Analyze All Span Types
Traces can combine server, client, internal, producer, and consumer spans, plus error status, events, and attributes, so you can follow one request across APIs, services, databases, and asynchronous components.
Best For
| Scenario | Why It Fits |
|---|---|
| Microservice request tracing | Captures the full request path across services so you can find the slow span and see whether the delay came from the application, a database call, or an external dependency |
| API latency analysis | Spans for inbound requests and downstream calls separate network latency from application processing time |
| Error investigation across services | Links related spans into one trace, so error status, events, and attributes isolate the failing component and root cause |
| Dependency visibility | Shows how services interact with databases, message queues, and third-party APIs, including retry patterns, timeouts, and queue delays |
| Release validation | Compare traces before and after a deployment to detect changes in span duration, error rate, or dependency behavior |
| OpenTelemetry standardization | One managed backend for services in different languages and runtimes, with consistent authentication |
| AI and LLM cost and latency transparency | GenAI semantic conventions record token consumption, prompts, and step costs as span attributes, visualized in Grafana |
Consider Alternatives If
| If You Need... | Consider | Why |
|---|---|---|
| Metrics collection and dashboards | Monitoring Service | Metrics use a separate observability product and ingestion model |
| Log collection and analysis | Logging Service | Logs use a separate observability product and ingestion model |
Key Considerations
Billing & Costs
- Main billing: Usage-based on two dimensions: trace ingestion volume (GB per billing period) and trace storage (GB retained during the retention period). No base fee and no minimum commitment.
- Ingestion rates (tiered by monthly volume): 0.25 EUR per GB up to 250 GB, 0.18 EUR per GB from 250 GB to 5 TB, 0.12 EUR per GB over 5 TB
- Storage rate: 0.015 EUR per GB per 30 days
- How usage is counted: Total size of received spans (name, timing, attributes, events, links, resource metadata), in decimal units (1 GB = 1,000,000,000 bytes)
- When idle: A pipeline that receives no traces creates no ingestion usage
- Not billed: Opening Grafana dashboards, searching or reading traces, retrieving pipeline metadata through the API, rotating a pipeline key, and creating, modifying, or deleting pipelines
- Cost control: Configure sampling in your applications to reduce the trace volume you send
Limitations
- A trace that exceeds the maximum trace size is rejected at ingestion and does not appear in Grafana. Reduce the number or size of spans and attributes, or split very large operations into separate traces.
- Sustained traffic above the ingestion rate limit results in discarded spans. OpenTelemetry SDKs and Collectors retry with backoff.
- When the live-trace limit is reached, new traces are refused until in-flight traces are completed and flushed.
- Spans that arrive after their trace was flushed to storage are stored as a separate, partial trace. Keep clock skew small and export spans promptly.
- Newly ingested traces can take a few seconds to become searchable with TraceQL. Lookups by exact trace ID become available sooner.
- TraceQL searches are bounded by a maximum query time range and a maximum number of returned results. Very broad searches may be truncated or rejected.
- Grafana access depends on at least one tracing pipeline on the contract in the selected region.
- Each pipeline uses one ingestion protocol. HTTP exporters must use the HTTP endpoint and gRPC exporters the gRPC endpoint.
- To fully remove a user's Grafana access, revoke the user's IAM access for the contract and remove the user from the Grafana organization.
Management Options
- Tracing Service API (v1) for creating, updating, and deleting pipelines and rotating keys
- SDKs listed in the IONOS CLOUD SDK and tools overview
- Grafana for trace search, visualization, and correlation