Feature Overview

Datadog Continuous Profiler gives you complete, always-on visibility into how your code behaves in production, with low overhead and minimal setup. Find the functions and threads that slow requests, diagnose memory leaks faster, and resolve performance issues with guided insights that help engineers at all levels take action confidently.

Resolve production performance issues quickly with always-on, code-level visibility

Never miss production issues originating from your code by continuously profiling every line across all hosts and containers
Pinpoint methods and code lines that are inefficient under production load, despite having performed well in pre-production environments
Correlate spans and traces with profiling data to reveal the true root cause of latency when traces lack enough detail

APM trace detail showing the Profiles tab with thread-level CPU activity correlated to an active span

Help every engineer uncover hidden issues in code behavior, stability, and performance

Use guided insights that surface the most relevant performance problems without requiring deep profiling expertise
Gain thread-level visibility into your requests to investigate parallelization, deadlocks, inefficient garbage collection, locks, errors, and more
Determine exactly what your slow methods are spending time on: CPU, locks, I/O, garbage collection, and more

Guided insight panel for a GC Pauses finding showing a severity summary and step-by-step remediation options

Track every deploy and eliminate performance regressions

Create custom dashboards to effectively track and analyze long-term, code-level metrics and detect performance variations in production
Compare code behavior and impact across hosts, versions, and time-ranges during canary, blue/green, or shadow deploys
Isolate the most inefficient functions to quickly troubleshoot faulty deployments and decide whether to roll back or ship a fix

Profiler Comparison view showing CPU time side-by-side between two service deploys with a regression highlighted

Quickly uncover the root cause of memory leaks and OOM crashes

Use the Memory Leaks flow to determine if your service is leaking and understand the scope of the issue
Use allocation profiling to spot excessive or unexpected memory usage and visualize long-lived objects
Compare memory behavior between deploys to identify when a leak was introduced and pinpoint the responsible code change