AWS performance testing starter: EC2, Fargate, Lambda

This is a basic starting point for conducting performance testing on AWS, for the compute services EC2, Fargate, and Lambda. The provided application is a small "Hello world" application written in Node (running on Fastify).

Tests are conducted using k6. They come configured for 500 virtual users, hitting an endpoint for 30 seconds. The test also checks to ensure that >99% of all hits are successes and that the p95 duration is under 200ms.

For the Lambda solution, there is full deployment automation in place using Serverless Framework. For EC2 and Fargate, there are steps and scripts provided for provisioning the infrastructure and building and pushing the container image that will be used.

Prerequisites

• Recent Node.js (ideally 18+) installed.
• Amazon Web Services (AWS) account with sufficient permissions so that you can deploy the necessary infrastructure.
• For the Lambda part, ideally some experience with Serverless Framework as that's what we will use to deploy the service and infrastructure.

Installation

Clone, fork, or download the repo as you normally would. Run npm install.

Configuration

Make sure to check these files out and change them if their use applies to your use case.

• scripts/push.sh
• scripts/ssh.sh
• serverless.yml

Once you have deployed some testing application(s), update the endpoints in scripts/test.js.

Commands

• npm run setup: Install k6 and Podman (assumes Mac)
• npm test: Run the k6 load test
• npm start: Run the Lambda application locally
• npm start:server: Run the containerized application locally
• npm run build: Package Lambda stack with Serverless Framework
• npm run build:image: Use Podman to build the image
• npm run push: Use Podman to build and push image to AWS ECR
• npm run deploy: Deploy Lambda stack to AWS with Serverless Framework
• npm run teardown: Remove Lambda stack from AWS

Setup

EC2

There is no automation provided beyond the scripts. Instead, follow the web console wizard and the below general steps.

Note that the described setup is not strictly equal to, for example, the Lambda setup unless you add (for example) HTTPS.

Basic instructions:

• Use Arm architecture
• Use the Amazon Linux AMI
• Use the t4g.micro (or other configuration)
• Use open inbound HTTPS traffic for the test traffic
• Allow SSH traffic from your IP so you can connect to the instance
• Add port 8080 (or whatever you are using) to the open/allowed ports in the security group
• Download the key pair so you can connect to the instance
• Connect to the instance; get the connection information in the EC2 instance view and update scripts/ssh.sh - run that script
• Inside the instance, copy paste the contents of scripts/ec2-setup.sh, it should end with the server being in a running state
• Find the public DNS address for the instance in the EC2 console
• You can now curl it, similar to curl ec2-{RANDOM}.{REGION}.compute.amazonaws.com:8080

Fargate

There is no automation provided beyond the scripts. Instead, follow the web console wizard and the below general steps.

You may want to follow this guide if you want a guided tour of the overall steps.

Note that the described setup is not strictly equal to, for example, the Lambda setup unless you add (for example) HTTPS.

Basic instructions:

• Create a task definition
  • Use Arm architecture
  • Set the configuration as desired
  • Point it to your ECR image
  • No need to fiddle with roles or other things if you don't want to
• Create a cluster
  • Point to the task definition
• Create a service in the cluster
  • Most of this should be simple
  • Add load balancers if desired
  • Add service scaling if desired
• Start the task
  • Once it has started, click the task to get the public IP

Lambda

Simply run npm run deploy (after having installed dependencies and being logged in through the CLI environment).

You'll get an endpoint you can use.

There are some configuration options in the serverless.yml if you want to set it up in other ways than the provided one.