Demo Lambda Insights

In this guide, you’ll learn how to provision multiple AWS Lambda functions with CloudFormation, invoke them to generate metrics, and explore those metrics in CloudWatch using Lambda Insights.

1. CloudFormation Template

Below is a CloudFormation YAML template that defines three Lambda functions:

Resources:
  HelloWorldFunction01:
    Type: AWS::Lambda::Function
    Properties:
      FunctionName: HelloWorldFunction01
      Runtime: python3.9
      Handler: index.handler
      Role: !GetAtt HelloWorldFunctionRole.Arn
      Timeout: 15
      Code:
        ZipFile: |
          import json
          def handler(event, context):
              print("HelloWorld from function 01")
              return {
                  'statusCode': 200,
                  'body': json.dumps('Hello from Lambda!')
              }

  HelloWorldFunction02:
    Type: AWS::Lambda::Function
    Properties:
      FunctionName: HelloWorldFunction02
      Runtime: python3.9
      Handler: index.handler
      Role: !GetAtt HelloWorldFunctionRole.Arn
      Timeout: 15
      Code:
        ZipFile: |
          import json
          def handler(event, context):
              print("HelloWorld from function 02")
              return {
                  'statusCode': 200,
                  'body': json.dumps('Hello from Lambda!')
              }

  HelloWorldFunction03:
    Type: AWS::Lambda::Function
    Properties:
      FunctionName: HelloWorldFunction03
      Runtime: python3.9
      Handler: index.handler
      Role: !GetAtt HelloWorldFunctionRole.Arn
      Timeout: 15
      Code:
        ZipFile: |
          import json
          import time
          def handler(event, context):
              print("HelloWorld from function 03")
              time.sleep(5)
              return {
                  'statusCode': 200,
                  'body': json.dumps('Hello from Lambda!')
              }

Function Summary

2. Invocation Script

Generate consistent metrics by invoking each function 50 times. Save the following as lambda_call.sh and make it executable.

#!/usr/bin/env bash

for i in {1..50}; do
  aws lambda invoke --function-name HelloWorldFunction01 --payload '{}' output01.txt
  aws lambda.invoke --function-name HelloWorldFunction02 --payload '{}' output02.txt
  aws lambda.invoke --function-name HelloWorldFunction03 --payload '{}' output03.txt
done

chmod +x lambda_call.sh

3. Deploying the Stack

1. Open the AWS Management Console and navigate to CloudFormation.
2. Choose Create stack > With new resources (standard).
3. Upload the YAML template and click Next.
4. Enter a stack name (e.g., lambda-insights-demo), acknowledge IAM capabilities, then click Next and Create stack.

Wait until the status reads CREATE_COMPLETE.

4. Verify Lambda Functions

In the Lambda console, confirm that all three functions (HelloWorldFunction01, HelloWorldFunction02, HelloWorldFunction03) are present and using Python 3.9.

5. Generate Load Using AWS CloudShell

1. Open AWS CloudShell from the console toolbar.

2. Upload or paste the lambda_call.sh script into your home directory.

3. Execute:

   ./lambda_call.sh
   

You’ll see output similar to:

{
  "StatusCode": 200,
  "ExecutedVersion": "$LATEST"
}
...

Notice that invocations for Function 03 take ~5 seconds due to the sleep.

6. Enabling Lambda Insights

By default, Lambda Insights is off. To enable it:

1. In the Lambda console, select each function and go to Configuration > Monitoring and operations.
2. Click Edit, enable Enhanced monitoring (Lambda Insights), and save.

Wait a few minutes for data to appear in CloudWatch.

7. Exploring Metrics in CloudWatch

Navigate to CloudWatch > Insights > Lambda Insights to view consolidated metrics across all functions.

To focus on a single function (e.g., HelloWorldFunction03), select it from the list:

8. Creating a Custom Dashboard

If you manage many functions, a dedicated CloudWatch dashboard helps.

1. In CloudWatch, go to Dashboards and click Create dashboard.
2. Add Lambda Insights widgets and select your functions.
3. Customize widgets for GB-s usage, duration, invocations, concurrent executions, and throttles.

Here’s an example with multiple functions:

Remember to disable Lambda Insights when not troubleshooting to avoid unnecessary costs.

9. Cleanup

1. In CloudShell, choose Actions > Restart CloudShell to stop any running processes.
2. In CloudFormation, select lambda-insights-demo and click Delete.
3. Confirm that the Lambda functions, logs, and IAM roles have been removed.

That completes this AWS Lambda Insights walkthrough. Happy monitoring!

References

• AWS Lambda Documentation
• AWS CloudFormation User Guide
• Amazon CloudWatch Documentation