knaufk / statefun-aws-lambda-demo

Serverless meets Apache Flink Stateful Functions on AWS

Stateful Functions is a project under the umbrella of Apache Flink. It makes it simple to build consistent, stateful applications at any scale.

This repository demonstrates how Stateful Functions complements FaaS:

• Stateful Functions makes it easy to compose larger applications out of individual functions. In this demo there are two functions: UserFn (implemented in Java) and GreeterFn (implemented in Python).
• Stateful Functions provides consistent fault-tolerant state and exactly-once messaging between functions. As a developer you don't need to worry about retries, network failures, etc.

Structure

• user-handler contains the implementation of the UserFn Lambda. This function is implemented in Java 11. The artifact is a JAR.
• greeter-handler contains the implementation of the GreeterFn Lambda. This is function is implemented in Python. The artifact is Docker image.
• generator contains the a small data generator that is deployed into the same Kubernetes cluster as Stateful Functions.
• k8s contains the Kubernetes manifests of the data generator and the Stateful Functions cluster. This also contains the module.yaml.
• terraform-function-repositories contains the terraform manifests for ECR and and an S3 Bucket to host the UserFn JAR.
• terraform-statefun contains the terraform manifests for everything else (EKS, Gateway, Lambda Functions).

Trying it out yourself

Prerequisites

• an AWS account
• terraform
• aws
• kubectl
• mvn
• docker
• envsubst

Setup

Step 1

Change the terraform backend to a pre-created bucket in main.tf and main.tf. Then, SET the variables in the Makefile to fit your environment (all except GATEWAY_ID)

make terraform-statefun

Step 2

Now, update the GATEWAY_ID variable in the Makefile to the value returned by make terraform-statefun.

make k8s

Trailing the Output

The date generator not only generates names to the names stream, but also reads the greetings stream and prints those records to STDOUT. So, you can use the following command to trail the greetings stream.

kubectl logs -n generator greeter-generator-<id>  -f 

Accessing the Flink UI

A Stateful Functions cluster boils down to a Flink Cluster running a specific Flink job. You can forward a local port to the Flink UI via the following command.

kubectl port-forward -n statefun services/statefun-master 8081