Documentation Grafana Cloud Cost Management and Billing Analyze and reduce costs Reduce costs Profiles costs

Reduce Grafana Cloud Profiles costs

Control your profiling costs by managing which services you profile, which profile types you collect, and how much label metadata you attach to your profiling data.

Grafana Cloud Profiles usage is measured in GB ingested per month. The main factors that affect your profiling costs are the number of services you profile, the number of profile types you enable per service, and the labels you attach to each profile. By reviewing these factors, you can reduce the volume of profiling data and lower your costs.

For more details about how profiles usage is calculated, refer to Understand your Grafana Cloud Profiles invoice.

Review your usage

You can review your profiles usage on the Billing and Usage dashboard. Configure usage alerts to notify your team when activity exceeds expected levels.

Reduce the number of services you profile

Profile only the services where profiling data provides actionable value. Services with stable, well-understood performance characteristics may not need continuous profiling.

Review the services sending profiling data and disable profiling for services that don’t require it. Focusing on critical or high-traffic services reduces data volume without losing visibility where it matters.

Select only the profile types you need

Each profile type, such as CPU, memory, goroutine, mutex, or block, generates its own data stream. Collecting all profile types for every service increases data volume.

The available profile types depend on the SDK you use to instrument your application. Refer to Profile types and instrumentation for the available profile types for each SDK.

Enable only the profile types relevant to the issues you’re investigating. For example, if you’re focused on CPU bottlenecks, you don’t need to collect memory or mutex profiles at the same time.

Manage labels

Labels add metadata to your profiles and increase the size of your profiling data. Adding unnecessary labels increases ingestion volume and costs.

Minimize the number of external labels you attach to each profile. If you use Grafana Alloy, you can use the pyroscope.relabel component to filter profiles or standardize the external labels before profiles are forwarded to Grafana Cloud. The pyroscope.relabel component only rewrites external labels, such as labels inferred by pyroscope.scrape or provided through the /ingest query parameter. It doesn’t modify labels embedded inside profile payloads like pprof sample labels.

Use Adaptive Profiles to reduce costs

You can use Adaptive Profiles to automatically adjust the detail and frequency of your profiling data to reduce costs during normal operation.

Adaptive Profiles is an intelligent sampling service that adjusts profiling resolution based on workload behavior. It captures detailed data when it matters most, such as after a deployment, and reduces data to a cost-effective baseline during normal operation.

Adaptive Profiles monitors your services at a low-cost baseline resolution and automatically increases the profiling detail when a triggering event occurs. This means you get high-fidelity data for investigating performance issues without paying for full-resolution profiling at all times.

Refer to the Adaptive Profiles documentation for more information.