Get started with Adaptive Traces

Enable Adaptive Traces with the recommended configuration in one click, then let recommendations refine your sampling over time.

Adaptive Traces helps you cost-effectively collect and retain the traces that matter most. It combines a volumetric sampling rate with automatic keep rules for errors, slow traces, and unique traces, plus always-on anomaly detection.

Start with the recommended configuration. It’s the right choice for almost everyone: it keeps your most valuable traces, significantly reduces your trace volume, and requires no ongoing maintenance. You can always add, edit, or remove policies later, but most customers never need to.

Enable Adaptive Traces

To get started with Adaptive Traces, complete the following steps.

Navigate to Adaptive Telemetry > Adaptive Traces.
Review the onboarding panel, which helps you choose which traces to keep as your starting point.
Choose which traces to keep.
Adaptive Traces provides three options. Each option sets a volumetric sampling rate, which distributes your sampling budget fairly across services and trace types, and keeps the traces that matter regardless of the rate:
- keep error traces
- keep slow traces
- keep unique traces
- keep traces related to detected anomalies
Recommended (5%) Start here. A balanced starting point with a 5% volumetric sampling rate.
If you want to keep more traces (25%) Choose this only if you have a known need for a larger sample of day-to-day traffic. It uses a 25% volumetric sampling rate with the same keep rules as the recommended option.
Custom Set your own volumetric sampling rate and choose which keep rules to include: keep error traces, keep slow traces, and keep unique traces. Traces related to detected anomalies are always kept. Only use this option if you already know your sampling requirements. You can achieve the same result later by editing the policies the recommended option creates.
Click Create policies.
Adaptive Traces creates volumetric, latency, status code, and diversity policies, based on the option and keep rules you chose.
No anomaly policy is created at this point. Anomaly detection is always on and creates temporary policies only when it detects unusual behavior, so anomaly related traces are kept without any standing policy.
Your sampling is now active. There’s nothing else you need to configure. You can edit or add policies at any time.

Apply recommendations as they appear

After you enable Adaptive Traces, it continues to analyze your traffic and suggests improvements as recommendations on the Overview page. Apply them when they appear; they’re the intended path from your starting configuration to an optimized one.

Volumetric sampling: The onboarding creates a volumetric policy for you. If you enabled Adaptive Traces before volumetric sampling was available and still have a single probabilistic policy, Adaptive Traces recommends upgrading it to a volumetric policy at the same percentage. Refer to Use volumetric sampling for more information.
Diversity sampling: The onboarding creates a diversity policy for you unless you chose the custom option and cleared keep unique traces. If you don’t have one, Adaptive Traces recommends it when it detects that rare or low-traffic traces may be underrepresented. Refer to Use diversity sampling for more information.
Anomaly detection: Enabled by default. When Adaptive Traces detects unusual behavior, such as a latency spike in a service, it temporarily keeps the relevant traces so you can investigate. Auto-applied anomaly recommendations appear in the Recommendations History on the Overview page. Refer to Investigate anomalies for more information.

Next steps

After your policies are created:

Open the Policies section and verify that the volumetric, latency, status code, and diversity policies you selected have been created.
Monitor your incoming traces to confirm your sampling level aligns with expectations. Refer to Monitor Adaptive Traces for more information.
If you need to go beyond the recommendations, for example to drop health check traffic, refer to Best practices for policies before you create custom policies.