Telemetry

Telemetry plays an important role in understanding and optimizing the performance of Java applications. YourKit Java Profiler captures a wide array of telemetry data, ranging from CPU usage to memory allocation, and I/O operations. This data is organized hierarchically in a tree structure and visualized through intuitive performance charts, with each telemetry metric presented in a dedicated chart for ease of analysis.

• How to control telemetry collection
  • Toolbar buttons
  • Profiler HTTP API
  • Profiler Java API
• Customization of the visible charts
• Working with telemetry charts
• Toolbar
• Changing the scale of the charts

How to control telemetry collection

Using the buttons in the profiler toolbar

1. Start : Clicking the start button initiates the collection of telemetry data. It's the first step you take when you want to begin monitoring the performance of your Java application.
2. Stop : The stop button is used to halt the collection of telemetry data. When you've gathered enough information, or you need to pause the profiling process for any reason, clicking stop will cease data collection, allowing you to analyze the collected data without accumulating any further information.
3. Reset : After stopping the telemetry collection, you might want to clear all the collected data to start a fresh session. The reset button clears all current telemetry data. This is particularly useful for performing multiple, separate profiling sessions without mixing data.

Programmatically via HTTP API

For users requiring remote control or automation within their profiling workflows, YourKit Java Profiler offers control over the telemetry collection through HTTP API endpoints. The documentation outlines startTelemetry, stopTelemetry and resetTelemetry methods available for this purpose.

Programmatically via Java API

The YourKit Java Profiler also provides a Java API for deep integration with Java applications. The com.yourkit.api.controller.v2.Controller includes methods for starting, stopping, and resetting the telemetry collection, similar to the toolbar buttons.

Customization of the visible charts

By utilizing the chart selector located on the left side of the interface, you can choose which metrics to monitor. This functionality supports you in identifying patterns and correlations across different performance metrics by stacking them together.

As a chart is selected in the hierarchical tree of metrics, the interface intelligently scrolls to bring the chosen chart into view, ensuring that you can quickly navigate through the wealth of information without losing track of the metric of interest.

Working with telemetry charts

Clicking on chart: Clicking on a chart reveals the call stacks of the Java threads running at that specific moment, providing a snapshot of the application's state.

Selecting the time interval: Analyzing performance over a time span is equally straightforward. By selecting a range within the telemetry data, profiler builds an estimated call tree and computes the CPU usage for that interval. This can be invaluable for pinpointing performance bottlenecks or understanding the behavior of the application during periods of high load.

Toolbar

• : Hide "empty" charts, i.e. the charts that have only zero values.
• : Smooth out short-term fluctuations on graphs. This is useful for visualizing of fast-changing data.
• , , : Quickly change the range of visible data.
• : Control auto-scrolling behavior of telemetry charts. When toggled, the graphs will be automatically scrolled to the right when new data arrives.

Changing the scale of the charts

Using the time range selector, which appears when the mouse pointer is over the chart, you can select the rage and scale of the visible data. You can also hold down the Ctrl key and use the mouse wheel to zoom the graphs.