Explore ways to use Java Runtimes with AWS Lambda with and without AWS Lambda SnapStart using different framewoks.

Architecture

Project Description

The code example include storing and retrieving product from the Amazon DynamoDB. I put Amazon API Gateway in front of my Lambdas.

I made all the test for the following use cases:

• Lambda function without SnapStart enabled
• Lambda function with SnapStart enabled but without usage of Priming
• Lambda function with SnapStart enabled but with usage of Priming (DynamoDB request invocation or proxing the whole web request)

I tested different impact of the Lambda cold and warm start of the following:

• Deployment package sizes
• Lambda memory settings
• Java compilation options
• Synchronous HTTP clients (Url Connection, Apache, synchronous AWS CRT)
• Asynchronous HTTP clients (Netty NIO, asynchronous AWS CRT)

To the explored AWS Lambda Java Runtimes belong:

• Java 11
• Java 17
• Java 21

To the explore frameworks belong :

• Micronaut
• Quarkus
• Spring Boot (including Spring Boot 3.2. version)

Installation and deployment

Clone git repository locally
git clone https://github.com/Vadym79/AWSLambdaJavaSnapStart.git

Compile and package the Java application with Maven from the root (where pom.xml is located) of the project
mvn clean package

Deploy your application with AWS SAM
sam deploy -g  

In order not to use AWS Lambda SnapStart comment both lines in the globals's section of the Lambda function.

Globals:
Function:
#SnapStart:
#ApplyOn: PublishedVersions

In order to user AWS Lambda SnapStart uncomment both lines above. For different Priming optimizations enabling of SnapStart is required

Further Readings

You can read my article series AWS Lambda SnapStart