Analysis reports

Analysis reports in Stackdriver Trace show you an overall view of the latency for requests to your application, or for a subset of the requests.

Daily reports

Trace automatically creates a daily report that compares the previous day's performance with the performance from the same day of the previous week. You can see this report in Trace > Analysis report:

Stackdriver Trace analysis report

The graph in the analysis report shows the distribution of requests in the report along a logarithmic axis of average latency times. The Latency panel below the graph view shows average latency times by percentage with links to traces for representative sample requests.

This analysis report is a simplified view of the full analysis reports that you can create in Trace.

Create a custom analysis report

You can create a custom analysis report using a single set of parameters, or generate a report over two sets of parameters, which allows you to compare two different sets of latency data and see additional information about your traces. Custom reports include:

  • Both the density and cumulative distribution graphs for latency data.
  • Bottlenecks, when detected, identify RPC calls that are significant contributors to latency.

To create a custom analysis report, you can:

Create a new analysis report

To create a new analysis report:

  1. Go to Stackdriver > Trace:

    Go to Trace

  2. On the Analysis reports tab, click New Report.

    Enter or select the following properties in the form that appears:

    Property Description
    Request filter The full URI, the URI prefix, or trace filter terms.

    For example, if a project root span is Recv, to include all requests, enter the root span name, Recv, or a prefix of the root span name. To restrict the report to traces with a URI prefixed by Recv/cart, enter that prefix into the filter.

    You can also filter traces by span attributes. For example, if you add /http/status_code:5, the report includes only those traces that have a span with the attribute key /http/status_code and attribute values prefixed by 5.
    HTTP Method HTTP method for the request. Default is All.
    HTTP Status HTTP status returned by the request. Default is All.
    Report Name Human-readable name for the report. If no name is specified for this optional field, Stackdriver Trace creates a name based on the specified request filter and the time at which the report was created.
    Module App Engine module (Java, Python, Go or PHP) that handled the request. Default is All.
    Version Version of the App Engine application that handled the request. Default is All.
    Time Range Time range during which the requests occurred. You can select Last Hour, Last 24 Hours, Last Week or Custom Range. If you select Custom, you must specify the start and end times for the range. Default is Custom.
  3. To create a report that compares two sets of data, select Create Comparison and specify the module, application version, and time range for the second data set.

  4. Click Submit.

Create a similar (duplicate) report

Stackdriver Trace also provides the option to Create a similar report, which duplicates the chosen report and allows you to edit one or more parameters. To create a report this way:

  1. View an existing analysis report.

  2. Click Create similar report.

  3. In the New report request window, edit one or more report parameters and click Submit.

View an analysis report

To view an analysis report:

  1. To to Stackdriver > Trace:

    Go to Trace

  2. Click the Analysis Reports tab.

    The analysis reports list shows both the system-generated daily report and any custom reports you have created.

  3. Click the name of a report.

    The analysis report appears.

View the overall request latency

To show a graphical distribution of requests in the report along a logarithmic axis of average latency times, click the Density distribution button in the analysis report.

Stackdriver Trace overall density distribution

If you created a comparison report, Stackdriver Trace overlays the data from the two data sets so that you can compare them and analyze any changes.

Stackdriver Trace overall comparison density

The Latency panel below the graph view shows average latency times by percentage with links to traces for representative sample requests. You can click a trace number to view its details in the trace details view.

Stackdriver Trace overall request latency

Stackdriver Trace attempts to identify performance bottlenecks and reports them, if found, in a Bottlenecks panel in the view. This panel highlights RPC calls that are significant contributors to latency in the requests and displays them by percentile with representative sample requests. You can then click a trace number to view its details.

In a report that compares two sets of latency data, the bottlenecks highlights the RPC calls that caused latency distribution changes between the two sets.

Stackdriver Trace request bottlenecks

View the cumulative request latency

To show the percentage of requests that have completed along a logarithmic axis of average latency times, click the Cumulative distribution button.

Stackdriver Trace cumulative latency

If you created a comparison report, Stackdriver Trace overlays the data from the two data sets so that you can compare them and analyze any changes.

Stackdriver Trace cumulative latency comparison

¿Te sirvió esta página? Envíanos tu opinión:

Enviar comentarios sobre…

Stackdriver Trace Documentation