Load Testing ​

Load Testing is a process of ensuring your API performs as expected during a typical and peak load without breaking or becoming unresponsive

The goal of API load testing is typically to identify potential bottlenecks or issues that may arise when the API is used at scale, and to provide information that can be used to optimize the API for better performance

1. Create a workflow

   loadtest.yml

   yaml

   version: "1.1"
   name: Load Test
   env:
     host: example.com
   tests:
     example:
       steps:
         - name: GET request
           http:
             url: https://${{env.host}}
             method: GET

2. Add the a load test configuration

   loadtest.yml

   yaml

   version: "1.1"
   name: Load Test
   env:
     host: example.com
   config:
     loadTest:
       phases:
         - duration: 2
           arrivalRate: 1
         - duration: 10
           arrivalRate: 2
   tests:
     example:
       steps:
         - name: GET request
           http:
             url: https://${{env.host}}
             method: GET

   In the following example, we configure a load test for a http service. Load-testing gRPC APIs is also possible

   Your load test can have multiple phases. Each phase has a duration, which defines how long a phase lasts and an arrivalRate, which sets how many virtual visitors (or requests) arrive per second during the duration of a phase

   WARNING

   By default, the test runner will skip subsequent steps after the previous ones have failed. Add continueOnFail configuration option to your workflow, if you want to continue test execution after a failed step

3. Add (optional) checks

   loadtest.yml

   yaml

   version: "1.1"
   name: Load Test
   env:
     host: example.com
   config:
     loadTest:
       phases:
         - duration: 2
           arrivalRate: 1
         - duration: 10
           arrivalRate: 2
       check:
         p99:
           - lte: 500
   tests:
     example:
       steps:
         - name: GET request
           http:
             url: https://${{env.host}}
             method: GET

   Checks are optional and allow you to validate that the resulted performance metrics match your expected values

   INFO

   See Workflow Syntax for available load test checks

4. Run the load test

   stepci run --loadtest loadtest.yml

   Result:

   response_time:
     min: ............................................................ 437
     max: ............................................................ 500
     avg: ............................................................ 453.3636363636364
     med: ............................................................ 446.5
     p95: ............................................................ 498
     p99: ............................................................ 500
   steps:
     failed: ......................................................... 0
     passed: ......................................................... 22
     skipped: ........................................................ 0
     errored: ........................................................ 0
     total: .......................................................... 22
   tests:
     failed: ......................................................... 0
     passed: ......................................................... 22
     total: .......................................................... 22
   checks
     p99: ............................................................ pass
   rps: .............................................................. 1.5705311250713878
   iterations: ....................................................... 22
   duration: ......................................................... 14008
   
   Workflow passed after 14.008s