The Trace Analysis page consolidates charts, filters, trace lists, application topology, and end-to-end aggregation for a single application. Use these tools to identify latency bottlenecks, locate errors, and map service dependencies.
Prerequisites
Before you begin, make sure that you have:
A Managed Service for OpenTelemetry instance with trace data reporting enabled
An application that appears on the Applications page
Open the Trace Analysis page
-
Log on to the Managed Service for OpenTelemetry console. In the left-side navigation pane, click Applications.
-
On the Applications page, select a region in the top navigation bar and click the name of the application.
On the Applications page, select a region in the top navigation bar and click the name of the application.
In the left-side navigation pane, click Trace Analysis.
Filter traces
Narrow down the trace list by time range, duration, errors, span names, servers, or tags.
In the upper-right corner of the Trace Analysis page, select a predefined or custom time range.
Click the search box in the upper part of the page. In the drop-down list, set one or more of the following filter conditions and click Search:
Filter How to use Time consumed Enter values in the Min and Max fields to filter traces by duration range Exception Select this option to show only traces that contain errors Span name Specify one or more span names to narrow results Machine/IP Specify one or more server IP addresses to narrow results Label (Tag) Specify one or more tags to narrow results You can combine multiple span names, IP addresses, and tags in a single search.
Enable full real-time analysis
By default, the Trace Analysis page displays sampled data summaries. After you specify a sampling ratio for span data, you can enable full real-time analysis to query the complete sampled dataset in real time rather than a statistical summary.
If no sampling ratio is specified or the sampling ratio is set to 100%, enabling full real-time analysis displays all collected span data within your Alibaba Cloud account.
On the Trace Analysis page, click Enable full real-time analysis in the upper-right corner.
Select a time range.
After you select a time range, the page displays the complete sampled data for that period and updates in real time.
Interpret the analysis charts
The Time change trend section contains three charts that help you spot patterns in span volume, duration, and errors over time.
| Chart | What it shows |
|---|---|
| Span count trend | Total spans and exceptional spans over time |
| Duration trend | Span duration over time |
| Duration distribution | How span durations are distributed across buckets |
To interact with the charts:
Hover over a data point to view detailed statistics.
Select a time range within a chart to zoom in on that period.
Click a legend entry to hide or show that data series.
Use the span name and Machine/IP lists
The Span name and Machine/IP lists show all spans and servers within your Alibaba Cloud account. Use these lists to quickly locate and filter by a specific span or server.
In either list:
Enter a keyword in the search box and click the search icon to filter entries.
Click Span quantity or Response time in the upper-right corner to sort entries in descending order.
Click a span name or IP address to apply it as a filter condition on the Trace Analysis page.
Group traces by field
Group traces by a specific attribute to compare behavior across span names, servers, or tags.
On the Trace Analysis page, select a field from the Group By drop-down list.
Click All Span or API call:
Option Scope All Span Groups traces based on the selected field across all spans API call Groups traces based on the selected field for the first span (the entry point) only Click a field value to view the matching traces.
View trace details
The Traces tab lists each trace with its ID, start time, span name, duration, and server IP address.
On the Traces tab:
Click a trace ID to open the waterfall chart for that trace.
Hover over a span name to view its duration, start time, tags, and log events.
View the application topology
The Application Topology tab maps dependencies between the current application and its upstream or downstream services. Use it to understand how requests flow between services and where latency concentrates.
For performance reasons, the topology aggregates request data from up to 5,000 traces.
Topology metrics
Each dependency line between two applications displays a label such as 90%/3x, which combines the request percentage and call multiplier. You can also view the duration ratio between applications.
| Metric | Definition | Example |
|---|---|---|
| Request percentage | The ratio of requests from Application A that involve a call to Application B, relative to the total requests received by A. Some requests are sent to Application B because Application A may filter all the requests based on an IF statement. | A receives 100 requests. 90 of those call B. Request percentage = 90%. |
| Call multiplier | The ratio of spans generated when A calls B, relative to the total spans received by A. | A receives 100 spans. Calls from A to B generate 300 spans. Call multiplier = 3x. |
| Combined label | Format: {request percentage}/{call multiplier}. | 90%/3x means 90% of requests in A call B, and each such request calls B three times on average. |
View end-to-end aggregation
The End-to-End Aggregation tab aggregates traces by span name and application name, providing a breakdown of request counts, durations, and error rates across the entire call chain. Use it to find which spans consume the most time or produce the most errors.
For performance reasons, end-to-end aggregation covers up to 5,000 traces.
Aggregation metrics
| Metric | Description | Example |
|---|---|---|
| Request Count / Request Percentage | The ratio of the current span's request count to the total request count. | With 100 total requests and a 10% request percentage, 10 requests reach the current span. |
| Span Amount / Request Multiplier | How many times each request calls the current span on average. Calculated as span count divided by request count. | 1.5x means each request calls this span 1.5 times on average. |
| Average Self Elapsed Time / Percentage | The exclusive duration of a span, excluding time spent in child spans. For asynchronous calls, child span durations are not deducted. | Span A takes 10 ms total. Its child Span B takes 8 ms. The self elapsed time for Span A is 2 ms. |
| Exception Count / Exception Percentage | The ratio of exceptional requests to total requests. The exception count may differ from the number of exceptional requests when the request multiplier exceeds 1, because a single exceptional request can contain multiple exceptions. | 3% means 3 out of every 100 requests are exceptional. |
On the End-to-End Aggregation tab:
Hover over a span name (shown in blue) to view related traces. Click a trace ID to open its waterfall chart.
Click an application name to go to the application overview.
View the waterfall chart of a trace
The waterfall chart displays the span name, timeline, application name, start time, server IP address, and status of each span in a trace. Use this view to trace the execution path of a request and pinpoint where latency or errors occur.
The IP Address column may display IP addresses or server names, depending on your settings on the Application Settings page. For more information, see Manage applications and tags.
In the Span Name column, hover over a span name to view its duration, start time, tags, and log events.
