Lambda Load Testing

This small program designed simplify load testing of a AWS Service. The project was inspired by the blog Managing AWS Lambda Function Concurrency and is powered by the HEY! project. The vision of the project is to reduce some of the setup time for lambda load testing and evade installiation frustrations by containing everything in Docker images.

For the step-by-step breakdown of request limits, throttling, and reserved concurrency check out the blog post The Everything Guide to Lambda Throttling, Reserved Concurrency, and Execution Limits

Pre-ReqS

Before we can begin testing your service there are a couple things that you need to have installed in your local enviorment. 1) The project compartmentalizes everything you need to deploy, test, and clean-up inside a docker container. Before begining make sure you have docker installed and can successfully run the Hello-World container.

2) When you first run the code it will execute a BASH script that extracts your AWS Access Key and AWS Secret Access Key from your enviorment variables. Don’t panic yet :) There’s no security threat because the container that uses those keys only exists on your local computer and will never be pushed to an external enviorment. That being said, you do need to have your credentials setup properly for everything to work. If you have not done so yet, install the AWS Command Line Interface and enter your AWS configuration.

$ pip3 install awscli --upgrade
$ aws configure

Setup

Now we can bring the project to your local computer and prepare it for the tests. 1) Clone the git repository.

$ git clone https://github.com/krapes/lambdaLoadTesting.git

2) Modify the serviceFiles directory with your service files.

Navigate to the serviceFiles directory. Currently it contains a dummy lambda and the serverless.yml

If you would like to test the load preformance of one lambda simply replace the code inside dummy.py with your function. (Make sure the file is still named dummy.py and that your lambda starts from main(event, context) )

If you have multiple files you can replace the whole folder with your service. The program will deploy your service, test all of the created endpoints, and then remove the service. It’s recommand that you change to name of the service inside your serverless.yml file so as to not affect your normal service.

Run

Now your project is ready to run. 1) Navigate to the top folder lambdaLoadTesting 2) In the command line execute:

$ bash run.sh

3) Wait for your results to appear in the terminal

Options

When executing from the command line there are a handful of optional parameters the user can pass.

Options:
  -n  Number of requests to run. Default is 200.
  -c  Number of requests to run concurrently. Total number of requests cannot
      be smaller than the concurrency level. Default is 50.
  -p  The AWS profile that will be used for AWS Access Key and Secret Key

Tender Loving Care

This project is far from complete. The ultimate vison would be a pytest like package that could be run with any Serverless Framework Service and produce a visual / analitical output of how it behaved under load. Taking it a step further would be to return warning statements for areas of weakness.

Any comments, suggestions, critiques, or recommendations are very much welcomed. Please feel free to open a pull request, create an issue, or email at kerri.rapes@gmail.com

I would love to hear what was liked, not liked, and given your specific background what would be changed. Peace, Kerri