Load Testing the Application
In order to see how our application withstands traffic peaks, we are going to use Locust to load test our 3 APIs simultaneously. For this test, we are to emulate the access of 250 concurrent users during 10 minutes spawning users at a rate of 10 users/sec
Export the stack output variables
To invoke our API’s, we first need to fetch the ApiUrl output variable that our CloudFormation stack gives us. So let us iterate through our stack and export all output variables as environment variables:
export ApiUrl=$(aws cloudformation describe-stacks --stack-name monitoring-app --output json | jq '.Stacks[].Outputs[] | select(.OutputKey=="ApiUrl") | .OutputValue' | sed -e 's/^"//' -e 's/"$//')
echo "export ApiUrl="$ApiUrl
Run the Load Test
In your Cloud9 environment, run the following:
cd ~/environment/serverless-observability-workshop/code/sample-app
locust -f locust-script.py -H ${ApiUrl} --headless -u 250 -r 10 -t 10m
Remember that this test is going to take 10 minutes to complete.
After its completion you should see you terminal with an output like the following:
Did you notice that the /POST operation presented no errors and maintained its latency relatively low while the two other /GET operations presented a really elevated error rate and overall latency?. In the next step, we will try to understand why.