Continuous Profiler

Datadog Continuous Profiler analyzes and compares code performance all the time and in any environment, including production, with negligible overhead. Quickly identify and optimize the most time- and resource-consuming parts in your application code in order to improve MTTR, enhance user experience, and reduce cloud provider costs.

Pinpoint hard to replicate production issues caused by inefficient code

Continuously profile each line of code in any environment without affecting application performance and user experience
Identify methods that are inefficient under production load, despite having performed well in pre-production environments
Optimize resource consumption and save on compute costs with code profiling aggregations across hosts, services, and versions

Method-level visibility into every request

Tie every distributed trace in APM to the performance of methods and threads that executed the request
Determine the root cause of slow requests with a breakdown of time spent by method on garbage collection, locks, and I/O
Resolve service bottlenecks and improve user experience by optimizing the methods that are causing high latency

Track every deploy and eliminate code performance regressions

Compare a variety of profile types to analyze the performance impact of new code during canary, blue/green, or shadow deploys
Isolate the most resource-heavy methods to quickly understand what is causing a spike in latency, CPU utilization, or memory allocation
Monitor code performance variations in production by applying code profiling metrics to your dashboards

Automated code profiling insights, leveraging years of runtime expertise

Derive actionable insights from an automatic heuristic analysis of the main problem areas in your code
Surface runtime performance problems such as deadlocked threads, inefficient garbage collection, and memory leaks
Apply suggested fixes to improve application performance without having prior experience in code profiling

We run Datadog Continuous Profiler on every service in our testing and production environment, which gives us the insight we need to reduce the time it takes to diagnose issues and fix performance bottlenecks.

Ian Schell

Site Reliability Architect at Cvent

Five minutes after setup, Continuous Profiler identified a significant amount of reflection being generated in an extremely business-critical service. The Profiler helped us quickly locate and eliminate this reflection across the service's codebase, resulting in a 40% improvement in message throughput.

Mike Clark

Staff Engineer at Olo

Datadog Continuous Profiler gives us unparalleled visibility into resource allocations in production with low overhead, and has become a crucial tool for optimizing CPU and memory performance at Faire.