Lambda Canary Based Deployment Using CodePipeline, CodeDeploy, and SAM

Canary Based Deployment

In this post, I will show you how we can deploy an AWS Lambda function using the “canary-based” deployment strategy.

The “canary-based” deployment helps to deploy a Lambda function gradually. With this approach, we can test the new version with some production traffic and roll back to the original if the new version did not work as expected. For example, we shift 10% of production traffic to the new version and verify the changes.

In this example, I use AWS SAM to create the Lambda function because it comes built-in with CodeDeploy to provide gradual Lambda deployments. With the help of AWS CodePipeline, we can create a fully automated CI/CD pipeline. I use AWS CDK for creating the infrastructure.

Steps

• Define the Lambda functions using AWS SAM
• Setup AWS CodePipeline and stages
• Running and testing the project

Define the Lambda Functions Using AWS SAM

As you can see in the below SAM template I have defined a Lambda function with the canary-based deployment configurations. In this example, I use LambdaCanary10Percent5Minutes as the deployment type. Also, I have configured a PreTrafic hook to verify the new version before traffic shifting.

LambdaCanary10Percent5Minutes: Shifts 10 percent of traffic in the first increment. The remaining 90 percent is deployed five minutes later.

PreTrafic: Lambda function that is run before traffic shifting.

AutoPublishAlias: The name of the Lambda alias. AWS SAM creates Lambda version and alias resources when this property is set.

You can clone the SAM project from this link

Main Lambda function

This is a simple function that returns the length of a string.