In the Google Cloud console, the Stackdriver Trace Overview window displays a summary of the latency data for your application and lets you investigate individual traces in more detail:
- Lists performance insights that can help you understand how to reduce your application's latency.
- Lists the most frequent application requests and RPC calls and their average latencies.
- Displays up to three daily analysis reports. Each report displays the latency data for the previous day for a single RPC endpoint. If data for an endpoint is available from seven days earlier, that earlier data is included in the graph for comparison purposes.
- A list of the most recent traces with their latency and the time they were made.
- A billing summary of your chargeable trace spans.
To get started using Stackdriver Trace, go to the Overview window:
This is the default view in Trace. To return to this window, click Overview in the navigation pane:
The Insights pane displays a list of performance insights for your application, if applicable:
The Insights pane highlights common problems in your application and can
help you reduce latency. For example, an insight might report that your
application is making too many consecutive calls to the Datastore
put() operation and that you might reduce latency if you batch the operations
To view detailed information about an insight, click the name of the insight. The Insights pane of the Trace details window describes the potential problem and recommends steps to resolve the problem:
The Recent traces pane displays the most recent traces for a project:
To view the details of a listed trace, click the URI to open the Trace details window. From the Trace details window, you can examine this trace in detail.
The Most frequent URIs pane displays a list of URIs for the most frequent requests to your application during the previous day, along with their average latency:
To browse and view detailed latency data for matching requests, click any URI to open the Trace list window. By default, the Request filter is populated and data from the previous 1 hour is displayed. You can change the filter and the time range to view additional requests:
For more information, go to Finding and viewing traces.
Frequent RPC calls
The Most frequent RPCs pane displays a list of the most frequent RPC calls made by your application during the last day, along with their average latency:
To browse latency data for requests where the matching RPC call was made, click an RPC name.
Chargeable trace spans
The Chargeable Trace Spans pane displays information related to your costs when using Trace:
- This month's trace spans ingested tells you the number of trace spans created and received by Trace after the first day of the current calendar month for your Google Cloud project.
- Last month's trace spans ingested tells you the total number of trace spans created and received by Trace in the last full calendar month for your Google Cloud project.
Using these values, you can estimate your costs of using Trace. For pricing information, go to Stackdriver pricing.
Daily analysis reports
The daily analysis reports pane displays up to three auto-generated reports. Each report displays the latency data for the previous day for a single RPC endpoint. If data for an endpoint is available from seven days earlier, that earlier data is included in the graph for comparison purposes. A report is generated for a RPC endpoint only if it is one of the three most frequent RPC endpoints, and only if there are at least 100 traces available.
If enough data isn't available to create at least one auto-generated report, Trace prompts you to create a custom analysis report.
The latency data can be displayed as a density distribution or as a cumulative distribution:
By default, the daily analysis report displays data as a density. The horizontal axis displays the latency using a logarithmic scale, and the vertical axis displays the percent of requests for each latency value. The density distribution lets you evaluate the consistency of latency values.
By clicking the Cumulative distribution button, the vertical axis changes to be the percent of requests whose latency is less than, or equal to the value latency value on the horizontal axis. This view lets you determine the percentage of requests whose latency is less than a particular value.
For more information, go to Creating and viewing analysis reports.