The real-time diagnosis feature is suitable for scenarios that need close monitoring of application performance and location of the cause of problems within a short time. This topic describes how to use the real-time diagnosis feature.
Background information
When you need to closely monitor the application performance for a short period of time, such as when releasing an application or performing stress tests on the application, you can use the real-time diagnosis feature. After real-time diagnosis is enabled, this feature continuously monitors the target application for five minutes and reports all the data of the traces during this period. Next, you can use the method stack waterfall chart and thread profiling to identify the causes of the exception based on the trace that shows performance problems.
Portal
Access the Real-time Diagnosis page as follows:
Enable and disable real-time diagnostics
The first time you access the Real-time Diagnosis page, real-time diagnostics is automatically enabled. To enable real-time diagnostics in other cases, click Enable real-time diagnosis in the upper-right corner.
Real-time diagnostics is automatically enabled for 5 minutes and then disabled. To disable real-time diagnostics, click Terminate Real-time Diagnosis in the upper-right corner.
View real-time monitoring data
In the Real-time Requests Distribution and Requests by Response Time sections, you can view the statistics of the last 1,000 requests captured as of the current point in time.

In the chart of the Real-time Requests Distribution section, select a time range. Data of the selected time range can be set as visible. The chart shows data only within this time range. Click Reset in the upper-right corner of the chart and the default view can be restored.

Filter monitoring data
You can filter request monitoring data displayed on the page by operation name or IP address.
View information of traces
On the Traces and Interfaces Aggregated tabs, you can view information of all traces captured in the corresponding period. Click a trace ID to access the Link Invocation page. Use the local method stack waterfall chart and thread profiling to identify the causes of exceptions.
