Continuous Profiling

Created:2024-09-23 Last Modified:2024-09-23

This document was translated by ChatGPT

#1. Continuous Profiling

For the working principle, see Core Features - Continuous Profiling Description

#1.1 Overview

Overview

Overview

  • ① Page Search: Search bar, search snapshots, and other functions. For usage details, please refer to the section Tracing - Resource Analysis
  • ② Quick Filter: Supports filtering by Application List and Profiling Type
    • Application List: Displays the list of reported services. The default display format is Service Name (Language Type)
    • Profiling Type: Displays the Profiling Types supported by the profiling method configured on the client
      • Select the Profiling Type supported by the current application. Different profiling types represent different semantics
  • ③ Display Switch: Switch the display mode of performance profiling data. Currently supports: Flame Graph, Table, and Both
    • Flame Graph: Displays the function call stack in the form of a flame graph
      • ⑥ Tip: Hover to view information about the Span in the flame graph
        • Function Type:
        • Span Name
        • Total Consumption: Percentage of the Span's total consumption relative to the root (first row of the flame graph)
        • Self Consumption: Percentage of the Span's self consumption relative to the root (first row of the flame graph)
      • Operation: Click to zoom in and view the call stack of the clicked Span; click on a blank area to return to the original state
    • Table: Displays Self Consumption and Total Consumption of functions in a list format
      • Default sorting is in descending order by Self Consumption
    • Both: View performance profiling data in both Flame Graph and Table formats simultaneously
      • Clicking a function in the table will highlight it in the flame graph
  • ④ Flame Graph Name Display: Choose whether to display the head or tail of the name in the flame graph
  • ⑤ Data Filter: Enter characters to filter data in the table