Load test regressions

You can specify integral metrics for each test in Load Testing. Use them to quickly assess the quality of the test or the service you are testing.

Employ the visualization tools on the page Regressions to build integral indicator versus time graphs for the tests you performed.

Your regression dashboard presents widgets with regression graphs, headings, and explanatory texts. Use it to visualize your load test data.

On the dashboard, you can set up regression charts with metrics for response time, number of testing threads, or RPS breaking points.

At the top of the dashboard, there is a timeline: use it to set the preferred regression graph interval. Above the timeline, there is the interval length selection panel with three vertical lines:

• The interval boundaries are colored blue.
• The current moment is colored red.

Regression typesRegression types

Response time: cumulative quantilesResponse time: cumulative quantiles

This chart shows how different-level quantiles (percentiles, to be more exact) depend on time.
An N-level quantile indicates the time that N% responses fall within. The 50th quantile is a median value and the 100th quantile is the maximum one.

Testing threads: quantityTesting threads: quantity

This chart shows how the number of testing threads depends on time.
A testing thread is a separate instance of the load generator. It usually equals the number of connections established between the agent and the service being tested.

Fault point: RPSFault point: RPS

This chart plots RPS breaking point versus time.

Thresholds for regression metricsThresholds for regression metrics

For each metric (line) on the regression dashboard charts, you can add a threshold value the excess of which should be considered abnormal. The chart areas where values exceed the thresholds are highlighted with the same color as the relevant metric line.

For individual tests, you can also use the check-regressions command in the Yandex Cloud CLI to check whether metric thresholds were exceeded. For more information, see Checking regression metrics for exceeded thresholds using the CLI.

See alsoSee also

Creating a regression dashboard
Checking regression metrics for exceeded thresholds
Managing a regression dashboard