Profiles

The Profiles tab embeds the Grafana Profiles (Pyroscope) view, pre-filtered to the selected service. It provides continuous profiling data to help identify performance bottlenecks at the code level.

Pre-applied filters

The Profiles tab automatically filters by the service’s namespace label.

Sub-tabs

Tab	Description
All services	Overview of profiling data across all services.
Profile types	Browse available profile types for the service.
Labels	Explore profiles grouped by label values.
Flame graph	View a flame graph visualization of the selected profile.
Diff flame graph	Compare flame graphs across two time periods to identify performance regressions.
Favorites	Quick access to starred profiles.

Profile types

Common profile types include:

CPU time consumed: identify which functions consume the most CPU.
Memory allocations: find memory-intensive code paths.
Block duration: identify contention points (Go).
Mutex duration: find lock contention (Go).
Goroutines: monitor goroutine counts (Go).

Filters and grouping

Service: select a specific service instance.
Profile type: select the metric to profile.
Filter by label values: narrow to specific instances or versions.
Group by labels: break down profiles by label dimensions.
Insights toggle: overlay knowledge graph insights on the profile data.

View modes

Choose how to display profiling data: Time series, Totals, Maxima, Histograms, Grid, or Rows. Enable Hide panels without data to focus on services with active profiling.

Flame graph and Exemplars

Click Flame graph to view the full flame graph for the selected profile and time range. Click Exemplars to correlate profile samples with specific trace exemplars.

Next steps

Use the Diff flame graph to compare before/after a deployment and identify performance regressions.
If no profile data appears, check the Quality report for instrumentation recommendations.
For complete profiling documentation, refer to Grafana Cloud Profiles.