Documentation Grafana Cloud Adaptive Telemetry Adaptive Profiles Introduction

Introduction to Adaptive Profiles

Learn more about the fundamentals and available features that help you optimize your use of Adaptive Profiles.

Fundamentals

The following concepts are key to your understanding of how Adaptive Profiles works.

• Insight

  An insight is an artifact automatically generated by the system to draw your attention to a specific, high-value performance issue.

• Resolution

  The level of detail being captured, primarily controlled by the sample rate and/or selection ratio.

• Baseline resolution

  The default, low-detail, and cost-effective sampling rate applied to all segments, ensuring that profiling is always active across all your services at a minimal cost.

• Increased resolution

  A temporary, high-detail sampling rate that is automatically triggered for a specific segment when the system detects a noteworthy event.

Prerequisites

Before you begin, ensure you have the following set up.

• You are sending profiling data to Grafana from the service you want to monitor.
• You have enabled the LLM plugin.
• You have enabled Adaptive Profiles.

Enable Adaptive Profiles

To get started with Adaptive Profiles, first enable the feature in Grafana Cloud.

1. Navigate to Adaptive Telemetry > Adaptive Profiles.

2. Enable Adaptive Profiles.

   Alternatively, you can enable it manually:

   • Go to Administration > Plugins and Data > Plugins.
   • Select Adaptive Profiles.
   • Open the Boosted Services tab and toggle Adaptive Profiles on.

3. (Optional but recommended) Enable the Grafana LLM plugin.

   Adaptive Profiles uses the LLM plugin to generate insights. Without it, data is still collected, but no insights appear on the Overview page.

How it works at a glance

Adaptive Profiles helps you find and fix performance issues faster by automatically capturing detailed data when it matters most, while lowering costs during normal operation. Once a service is enabled, Adaptive Profiles works automatically in the background.

1. Baseline monitoring

   The system monitors each of your service’s segments at a cost-effective baseline resolution. This provides an always-on, complete historical record of performance without high costs.

2. Sampling adjustment

   The system constantly analyzes your incoming profiles. When a triggering event occurs, such as a version change in a specific service, Adaptive Profiles automatically increases the profiling resolution for that service for a temporary period.

   Detailed profiles are collected for a short period (for example, 5–10 minutes), providing rich data for analysis during the critical post-deployment window.

   You can also manually increase the profiling resolution for a desired length of time.

3. Insight generation

   The high-resolution data is collected and then automatically fed into your Insights overview. If a valuable insight is found, the system creates an event to draw your attention to the specific performance issue. If no significant insight is found, the resolution returns to the baseline.

Your workflow

When an insight is generated, it’s your turn to act.

1. Review the insight

   Open insights from the Overview page to immediately see a summary, recommended improvements, and an analysis of the problematic functions.

2. Explore the data

   From the Insight, you can go to Drilldown for a more detailed flame graph to get more context and validate the findings.

3. Resolve the issue

   Based on the recommendations and your analysis, you can decide whether to make a change in your code.

4. Dismiss the insight

   Once you have addressed the problem or decided no action is needed, you can dismiss the insight, which moves it to your history for future reference.

Deactivate Adaptive Profiles

To deactivate Adaptive Profiles, navigate to Administration > Plugins and Data > Plugins > Adaptive Profiles and select the Boosted Services tab. Click the toggle to deactivate. Services are now be ingested and stored at 100% sample resolution.