AWS Cloud Path Week 18: Building a Serverless Coffee Shop Backend

In this session of AWS Cloud Path, we try our hands at building a serverless backend for a coffee shop ordering system using AWS Step Functions, Lambda, DynamoDB, and EventBridge. The system handles order processing, barista notifications, and order status updates in a completely serverless architecture. Check out the workshop here: Serverlesspresso Workshop

Missed the session? Catch up here:

Prerequisites

• AWS Account with appropriate permissions
• Basic understanding of serverless architecture
• Familiarity with AWS services

The Coffee Shop System Overview

The application consists of three front-end components:

1. Overhead monitor display
2. Barista application
3. Customer ordering application

The backend system needs to handle requests from all these front-ends while managing different data payloads for each interface.

Key Features

• QR code-based ordering system with limits (10 drinks/5 minutes)
• Barista queue management (maximum 20 drinks at a time)
• Real-time order status updates
• Shop status management (open/closed)

Building the Workflow with AWS Step Functions

1. Initial Setup

The workflow begins by checking two crucial conditions:

• If the shop is open
• If the barista queue has capacity

We implement this using AWS Step Functions state machine with the following components:

{
  "Comment": "Coffee Shop Order Processing Workflow",
  "StartAt": "CheckShopStatus",
  "States": {
    "CheckShopStatus": {
      "Type": "Task",
      "Resource": "arn:aws:states:::dynamodb:getItem",
      "Parameters": {
        "TableName": "serverless-coffee-config",
        "Key": {
          "pk": {"S": "config"},
          "sk": {"S": "shop"}
        }
      },
      "Next": "IsShopOpen"
    }
    // Additional states follow
  }
}

2. Shop Status Verification

The workflow first queries DynamoDB to check if the shop is open. This is implemented as a choice state that branches based on the shop's status:

• If shop is closed: Emit event via EventBridge
• If shop is open: Proceed to capacity check

3. Capacity Management

The system checks if the barista queue has capacity by listing current Step Functions executions:

• Maximum limit: 20 drinks in queue
• Uses Step Functions List Executions API
• Implements capacity verification before accepting new orders

4. Order Number Generation

When an order is accepted, we generate a unique order number using DynamoDB:

{
  "TableName": "counting-table",
  "Key": {
    "id": {"S": "orderId"}
  },
  "UpdateExpression": "SET id_value = id_value + :inc",
  "ExpressionAttributeValues": {
    ":inc": {"N": "1"}
  },
  "ReturnValues": "ALL_NEW"
}

5. Event-Driven Communication

The workflow uses EventBridge for event-driven communication:

• Order started events
• Shop status updates
• Order completion notifications
• Error handling events

6. Timeout Management

The system implements two crucial timeout scenarios:

• Customer timeout: 900 seconds (15 minutes) for order placement
• Barista timeout: 900 seconds for order preparation

Each timeout is handled with appropriate error states and event emissions.

Error Handling

The workflow implements comprehensive error handling:

• Customer timeouts
• Barista timeouts
• Shop closure scenarios
• Queue capacity issues

Each error case emits specific events that can be handled by other components of the system.

Testing the Workflow

To test callbacks in the workflow:

# For successful completion
aws stepfunctions send-task-success --task-token <TASK_TOKEN> --task-output "{}"

# For failure scenarios
aws stepfunctions send-task-failure --task-token <TASK_TOKEN> --error "Error.Type" --cause "Error description"

Next Steps

In the next session, we'll focus on:

• Connecting front-end applications
• Testing the complete system

The completed backend will provide a robust foundation for the coffee shop ordering system, handling everything from order placement to completion in a serverless architecture.

Remember to check your AWS free tier status before deploying, as the total cost for these services should be under $1 with free tier benefits.

Happy coding! ☕️