Menu
Grafana Cloud
Application Observability Pipeline
Applications may send telemetry data directly to the Grafana Cloud OTLP gateway, but it is recommended to run a Grafana Agent or OpenTelemetry Collector for several reasons:
- Reliability: The Agent and Collector buffer data during issues with the remote endpoint, ensuring transmission after recovery. Without the Agent, data loss occurs.
- Cost control: The Agent aggregates and drops telemetry data as necessary. This process allows for the forwarding of only relevant data, reducing costs.
- Flexibility: Using an Agent simplifies changes to the telemetry destination. Users update the Agent configuration rather than restarting applications, which is necessary for direct OTLP writing.
Two options exist: Grafana Agent or the OpenTelemetry Collector. Both work with Application Observability, but Grafana support only extends to Grafana Agent.
Grafana Agent
Consult the Grafana Agent export documentation for the recommended and supported way to send telemetry data for infrastructure and applications to Grafana Cloud.
OpenTelemetry Collector
Alternatively, consult the Grafana OpenTelemetry Collector export documentation to use the vendor-neutral and community supported OpenTelemetry Collector to send telemetry data to Grafana Cloud.
Was this page helpful?
Related resources from Grafana Labs
Additional helpful documentation, links, and articles:

Getting started with your metrics, logs, and traces in Grafana Cloud
Learn to use the best open source observability software (Grafana, Prometheus, Loki, and Tempo) without the overhead of managing and scaling your own stack.

Kubernetes monitoring, out-of-the-box with Grafana Cloud
In this webinar you’ll learn how Grafana offers developers and SREs a simple and quick-to-value solution for monitoring their Kubernetes infrastructure.

Unify your data with Grafana plugins: Datadog, Splunk, MongoDB, and more
In this webinar, learn how to leverage Grafana's plugin ecosystem for access to 80+ data sources, including plugins for Datadog, Splunk, MongoDB, and more.