All Products
Search
Document Center

:Continuous profiling

Last Updated:Jun 20, 2026

Continuous profiling helps you identify CPU-related performance bottlenecks in your applications. It provides detailed statistics broken down by method name, class name, and line number. Use this data to optimize your code, reduce latency, increase throughput, and lower costs. This topic describes how to view continuous profiling data.

Continuous profiling data

  1. Log on to the ARMS console. In the left-side navigation pane, choose Application Monitoring eBPF > Application List.

  2. In the top navigation bar, select a region.

  3. On the Application List page, click the name of the application.

  4. In the top navigation bar, choose application diagnostics > continuous profiling.

  5. Select a profiling type to view the CPU profiling data for a single instance.

    The continuous profiling feature is built on eBPF, which allows it to operate without instrumenting your code. For languages with debug symbols, such as Go, profiling data shows specific method calls. If you see an unknown method, it means the kernel could not find the corresponding symbol table for that method.

    At the top of the page, you can select a time window (1 minute/5 minutes/15 minutes) and switch between view modes using the Table, Both, and flame graph buttons. The toolbar provides options such as Head first, Reset View, and Focus on subtree.

    • The Self column shows the time or resources consumed by a method's own execution, excluding the time or resources consumed by its child method calls. This data helps you identify which methods spend excessive time or resources on their own execution.

    • The Total column includes the time or resources consumed by the method itself and by all of its child method calls. This helps you understand which methods contribute the most time or resources across the entire call stack.

    To investigate specific hotspot code, focus on the Self column or look for wider bars at the bottom of the flame graph on the right. A wide bar represents the root cause of high resource consumption in the upper call stack and typically indicates a system performance bottleneck.

Contact us

If you have any questions about using Application Monitoring eBPF Edition, join the DingTalk group (ID: 35568145) to obtain technical support.

Related topics

To learn more about troubleshooting issues related to CPU, memory, and I/O in Java applications, use the continuous profiling feature. For more information, see Enable continuous profiling.