Serverless Shopping Cart Microservice

This application is a sample application to demonstrate how you could implement a shopping cart microservice using serverless technologies on AWS. The backend is built as a REST API interface, making use of Amazon API Gateway, AWS Lambda, Amazon Cognito, and Amazon DynamoDB. The frontend is a Vue.js application using the AWS Amplify SDK for authentication and communication with the API.

To assist in demonstrating the functionality, a bare bones mock "products" service has also been included. Since the authentication parts are likely to be shared between components, there is a separate template for it. The front-end doesn't make any real payment integration at this time.

Architecture & Design

Design Notes

Before building the application, I set some requirements on how the cart should behave:

• Users should be able to add items to the cart without logging in (an "anonymous cart"), and that cart should persist across browser restarts etc.
• When logging in, if there were products in an anonymous cart, they should be added to the user's cart from any previous logged in sessions.
• When logging out, the anonymous cart should not have products in it any longer.
• Items in an anonymous cart should be removed after a period of time, and items in a logged in cart should persist for a longer period.
• Admin users to be able to get an aggregated view of the total number of each product in users' carts at any time.

Cart Migration

When an item is added to the cart, an item is written in DynamoDB with an identifier which matches a randomly generated (uuid) cookie which is set in the browser. This allows a user to add items to cart and come back to the page later without losing the items they have added. When the user logs in, these items will be removed, and replaced with items with a user id as the pk. If the user already had that product in their cart from a previous logged in session, the quantities would be summed. Because we don't need the deletion of old items to happen immediately as part of a synchronous workflow, we put messages onto an SQS queue, which triggers a worker function to delete the messages.

To expire items from users' shopping carts, DynamoDB's native functionality is used where a TTL is written along with the item, after which the item should be removed. In this implementation, the TTL defaults to 1 day for anonymous carts, and 7 days for logged in carts.

Aggregated View of Products in Carts

It would be possible to scan our entire DynamoDB table and sum up the quantities of all the products, but this will be expensive past a certain scale. Instead, we can calculate the total as a running process, and keep track of the total amount.

When an item is added, deleted or updated in DynamoDB, an event is put onto DynamoDB Streams, which in turn triggers a Lambda function. This function calculates the change in total quantity for each product in users' carts, and writes the quantity back to DynamoDB. The Lambda function is configured so that it will run after either 60 seconds pass, or 100 new events are on the stream. This would enable an admin user to get real time data about the popular products, which could in turn help anticipate inventory. In this implementation, the API is exposed without authentication to demonstrate the functionality.

Api Design

Shopping Cart Service

GET
/cart
Retrieves the shopping cart for a user who is either anonymous or logged in.

POST
/cart
Accepts a product id and quantity as json. Adds specified quantity of an item to cart.

/cart/migrate
Called after logging in - migrates items in an anonymous user's cart to belong to their logged in user. If you already have a cart on your logged in user, your "anonymous cart" will be merged with it when you log in.

/cart/checkout
Currently just empties cart.

PUT
/cart/{product-id}
Accepts a product id and quantity as json. Updates quantity of given item to provided quantity.

GET
/cart/{product-id}/total
Returns the total amount of a given product across all carts. This API is not used by the frontend but can be manually called to test.

Product Mock Service

GET
/product
Returns details for all products.

/product/{product_id}
Returns details for a single product.

Running the Example

Requirements

python >= 3.8.0 boto3 SAM CLI, >= version 0.50.0
AWS CLI
yarn

Setup steps

Fork the github repo, then clone your fork locally: git clone https://github.com/<your-github-username>/aws-serverless-shopping-cart && cd aws-serverless-shopping-cart

If you wish to use a named profile for your AWS credentials, you can set the environment variable AWS_PROFILE before running the below commands. For a profile named "development": export AWS_PROFILE=development.

You now have 2 options - you can deploy the backend and run the frontend locally, or you can deploy the whole project using the AWS Amplify console.

Option 1 - Deploy backend and run frontend locally

Deploy the Backend

An S3 bucket will be automatically created for you which will be used for deploying source code to AWS. If you wish to use an existing bucket instead, you can manually set the S3_BUCKET environment variable to the name of your bucket.

Build and deploy the resources:

make backend  # Creates S3 bucket if not existing already, then deploys CloudFormation stacks for authentication, a 
product mock service and the shopping cart service.  

Run the Frontend Locally

Start the frontend locally:

make frontend-serve  # Retrieves backend config from ssm parameter store to a .env file, then starts service.  

Once the service is running, you can access the frontend on http://localhost:8080/ and start adding items to your cart. You can create an account by clicking on "Sign In" then "Create Account". Be sure to use a valid email address as you'll need to retrieve the verification code.

Note: CORS headers on the backend service default to allowing http://localhost:8080/. You will see CORS errors if you access the frontend using the ip (http://127.0.0.1:8080/), or using a port other than 8080.

Clean Up

Delete the CloudFormation stacks created by this project:

make backend-delete

Option 2 - Automatically deploy backend and frontend using Amplify Console

1. Use 1-click deployment button above, and continue by clicking "Connect to Github"
2. If you don't have an IAM Service Role with admin permissions, select "Create new role". Otherwise proceed to step 5)
3. Select "Amplify" from the drop-down, and select "Amplify - Backend Deployment", then click "Next".
4. Click "Next" again, then give the role a name and click "Create role"
5. In the Amplify console and select the role you created, then click "Save and deploy"
6. Amplify Console will fork this repository into your GitHub account and deploy it for you
7. You should now be able to see your app being deployed in the Amplify Console
8. Within your new app in Amplify Console, wait for deployment to complete (this should take approximately 12 minutes for the first deploy)

Clean Up

Delete the CloudFormation stacks created by this project. There are 3 of them, with names starting with "aws-serverless-shopping-cart-".

License

This library is licensed under the MIT-0 License. See the LICENSE file.