Lambdas are used frequently - microservices, data pipelines, DevOps automation projects, cloud automation and many more. The selection of the lambda deployment mechanism is important. It can speed up development or prototyping work or slow it down significantly.

This project is showing few ways to deploy lambda. Each way is a bit different and can be used for different purposes. Enjoy!

Lambdas are deployed based on 2 types of artifacts:

• Zip packages (direct upload or via S3)
• Docker images (from ECR using AWS provided base image or custom one)

This depends on the size of lambda - generally, it is better to use Zip packages in small lambdas. There is a limit for using Zip packages - over 250MB (including dependencies, all uncompressed) you can use only Docker Images. If an application is larger than 50MB - the direct upload will not work. Use the S3 bucket as artefact storage for a deployment.

Common for all lambdas invoke script was placed in folder test To test lambdas execution create a bucket, place a file, add metadata and based on event.json.tpl create event.json with filled values. Lambda invoke script is in invoke.sh

Application is placed in code folder and will be used by all approaches. Application-based on file in S3 bucket returns its metadata using boto3 library. One of the common practises while deploying lambda is to store some common files and libraries in layer shared between different lambda functions. Code for such layer was placed in a layer folder.

Each approach contains:

• deploy, destroy scripts or README commands to run (i.e. for terraform)
• README with description of PRO/CONS and challenges

• manual - bash scripts with AWS CLI
• terraform - terraform scripts
• serverless - serverless framework configuration files
• sam - AWS SAM deployment

• Zappa
• ClaudiaJS
• Heroku
• AWS CDK (Cloud Development Kit)
• Up
• Architect

After exploring 4 different approaches to lambda deployment each has its own merits and drivers to consider.

This is a great approach for educational purposes. It can clearly show the shortcomings and required feature to look for when selecting tools. People who implemented this basic approach by themselves understand the lambda deployment mechanism better.

Recommended: education only

This approach has few versions of implementation. The one implemented in this repository was selected because of its short form and coverage of the most important features. It does not support debugging, logging or local execution - features that might come in handy. Terraform focus mainly on the deployment part.

Recommended:

• selected IaC is Terraform
• multi-cloud environment
• do not care about lambda development helpers - AWS CLI can be a replacement for most of them

This approach uses tool created especially for developing serverless application. Supports multiple providers to deploy serverless solutions in multi-cloud environments. Works based on AWS CloudFormation stack (for AWS provider) and provides nice integration with multiple triggers (like API Gateway). Supports development tools like local and remote execution, logging. In "Pro" version gives out-of-the-box CICD pipeline. Can be mixed with IaC tools like Terraform.

Recommended:

• easy integration with different triggers
• there are plugins available for all needed extensions
• focus on a tool to develop lambdas (incl. local execution)
• multi-cloud environments
• CloudFormation as IaC tool

The approach focused only on AWS serverless applications. An application consists of many lambdas and resources around them (like S3 buckets, API Gateways). CLI generates a whole project including tests based on a given template, which gives way to standardize serverless application development. CLI supports nicely code development, deployment and troubleshooting. During deployment and writing different lambdas there was no need to jump back to the AWS console - all needed data can be easily extracted from AWS SAM CLI.

Recommended:

• deployment of multiple lambdas as a part of a bigger application
• leverage of SAM application repository (public and private)
• focus on a tool to develop lambdas (incl. local execution)
• CloudFormation as IaC tool

• https://docs.aws.amazon.com/cli/latest/reference/
• https://docs.aws.amazon.com/lambda/latest/dg/lambda-python.html
• https://www.serverless.com/learn/comparisons/
• https://aws.amazon.com/blogs/opensource/24-open-source-tools-for-the-serverless-developer-part-1/
• https://registry.terraform.io/modules/terraform-aws-modules/lambda/aws/latest
• https://docs.aws.amazon.com/serverless-application-model/latest/developerguide/what-is-sam.html
• https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/stacks.html
• https://www.serverless.com/examples/
• https://www.serverless.com/framework/docs/