Advanced15 min read

Serverless Architecture

Build serverless applications with Lambda, API Gateway, event-driven patterns, and automatic scaling without infrastructure management.

Serverless Architecture Patterns

Serverless eliminates server management — AWS handles provisioning, patching, and scaling. Pay per invocation and compute time. Ideal for variable traffic, event processing, and microservices.

Core services: Lambda (compute), API Gateway (HTTP), DynamoDB (data), S3 (storage), SQS/SNS (messaging), EventBridge (events), Step Functions (orchestration).

No idle costs — pay only when code executes
Automatic scaling from zero to thousands of concurrent executions
15-minute maximum execution time for Lambda

# Serverless stack
# API Gateway → Lambda → DynamoDB
# S3 upload → Lambda (process) → DynamoDB
# EventBridge schedule → Lambda (batch job)
# SQS queue → Lambda (worker) → S3

API Gateway

API Gateway creates REST, HTTP, and WebSocket APIs. HTTP APIs are simpler and cheaper for Lambda integrations. REST APIs offer more features: request validation, API keys, usage plans, and caching.

Configure routes, integrations (Lambda, HTTP proxy), authorizers (Cognito, Lambda, IAM), and CORS. Deploy to stages (dev, staging, prod) with stage variables.

# Serverless Framework / SAM template excerpt
Resources:
  ApiFunction:
    Type: AWS::Serverless::Function
    Properties:
      Handler: index.handler
      Runtime: nodejs20.x
      Events:
        ApiEvent:
          Type: HttpApi
          Properties:
            Path: /users
            Method: GET

Event-Driven Processing

Connect services with events instead of direct calls. S3 uploads trigger Lambda for image processing. DynamoDB Streams trigger Lambda for change data capture. EventBridge routes events between services with rules and targets.

Event-driven architectures are loosely coupled, scalable, and resilient. Each service reacts to events independently.

# S3 event notification → Lambda
aws s3api put-bucket-notification-configuration \
  --bucket uploads \
  --notification-configuration '{
    "LambdaFunctionConfigurations": [{
      "LambdaFunctionArn": "arn:aws:lambda:...:function:process-upload",
      "Events": ["s3:ObjectCreated:*"],
      "Filter": {"Key": {"FilterRules": [{"Name": "suffix", "Value": ".jpg"}]}}
    }]
  }'

Step Functions

Step Functions orchestrate multi-step workflows with visual state machines. Define workflows in Amazon States Language (JSON/YAML). Built-in error handling, retries, and parallel execution.

Use for: order processing, data pipelines, human approval workflows, and microservice orchestration. Standard workflows for long-running tasks, Express for high-volume short workflows.

{
  "StartAt": "ValidateOrder",
  "States": {
    "ValidateOrder": {
      "Type": "Task",
      "Resource": "arn:aws:lambda:...:function:validate",
      "Next": "ProcessPayment",
      "Retry": [{"ErrorEquals": ["States.TaskFailed"], "MaxAttempts": 3}]
    },
    "ProcessPayment": {
      "Type": "Task",
      "Resource": "arn:aws:lambda:...:function:payment",
      "End": true
    }
  }
}

Serverless Best Practices

Keep functions small and focused — one function per responsibility. Minimize cold start impact with provisioned concurrency for latency-sensitive paths. Use Lambda layers for shared dependencies.

Implement idempotency for event processing — events may be delivered more than once. Use DynamoDB conditional writes or idempotency keys. Monitor with X-Ray and CloudWatch.

# Optimize Lambda package size
# - Use esbuild/webpack tree shaking
# - Lambda layers for shared deps (aws-sdk is included in runtime)
# - ARM64 (Graviton) for 20% better price-performance

# Provisioned concurrency for critical paths
aws lambda put-provisioned-concurrency-config \
  --function-name my-api \
  --provisioned-concurrent-executions 10

Serverless Architecture