One of the most tested patterns in SAA-C03 is: "How do you make this tightly coupled system scalable?" Loose coupling is a core principle of modern cloud architecture.
Why Do We Need Loose Coupling?
Consider a restaurant. If every server runs directly to the kitchen to explain each order, chaos erupts during rush hour. Orders pile up, servers wait, and one slow cook backs up the whole system.
Now add an order slip system. Servers drop the slip and move on to the next table. The kitchen works through slips at its own pace. Both sides operate independently. That is loose coupling.
Three key benefits of loose coupling: Each component scales independently If one part fails, the rest keeps running Each part can be deployed and updated independently
Amazon SQS — The Order Slip System
SQS (Simple Queue Service) is AWS's message queue service. It acts as a buffer between producers (senders) and consumers (receivers).
SQS Standard queues have virtually unlimited throughput. Message ordering is not guaranteed, but they can handle millions of messages per second. Right for most workloads.
SQS FIFO queues guarantee exact ordering and deduplication. Use them when order matters — "order 1 must be processed before order 2." Throughput is capped at 3,000 messages per second (with batching).
The Dead Letter Queue (DLQ) moves messages that fail processing into a separate queue. You can analyze and reprocess failed messages. Essential for error isolation and debugging.
Visibility timeout is an important concept. When a consumer retrieves a message, it becomes invisible to other consumers for a set time. If processing completes, the consumer deletes it. If processing fails, the message reappears in the queue after the timeout expires.
Amazon SNS — The Public Announcement Speaker
SNS (Simple Notification Service) is a fan-out service that delivers one message to multiple subscribers simultaneously. Think of a school PA system broadcasting to the entire building at once.
Publish a message to an SNS topic, and every subscriber receives it simultaneously. Subscribers can be SQS queues, Lambda functions, HTTP endpoints, email addresses, and more.
The SNS + SQS fan-out pattern appears very frequently on the exam. Subscribe multiple SQS queues to a single SNS topic, and one event gets delivered to multiple processing pipelines at the same time. For example, an order completion event can simultaneously reach a payment processing queue, an inventory update queue, and an email notification queue.
!Amazon SQS versus SNS
Amazon EventBridge — Central Hub for Event-Driven Architecture
EventBridge is an event bus that collects events, filters them by rules, and routes them to the right destination.
How it differs from SQS and SNS: EventBridge looks at the content of events to decide where to route them. You can build complex event-driven workflows like "when an EC2 instance terminates, trigger this Lambda" or "when a file is uploaded to S3, start this Step Functions workflow."
EventBridge can receive events from AWS services and also from third-party SaaS applications.
Amazon API Gateway — The Gatekeeper for Serverless APIs
API Gateway creates and manages REST, HTTP, and WebSocket APIs. Combined with Lambda, you get a fully serverless API — not a single server to operate.
Features API Gateway provides: Authentication and authorization (IAM, Cognito, Lambda Authorizer) Request throttling and quota management SSL/TLS termination Response caching to reduce backend load
Exam hint: "serverless REST API" or "expose Lambda over HTTP" — choose API Gateway.
Container Orchestration
In a microservices architecture, containers are the standard way to package each service.
| Service | Description | Analogy | |---------|-------------|---------| | Amazon ECS | AWS-native container orchestration | AWS's own container management platform | | Amazon EKS | Kubernetes-based container management | Industry-standard Kubernetes managed by AWS | | AWS Fargate | Run containers without managing servers | The serverless version of containers |
With Fargate, you never manage EC2 instances yourself. Say "run this container" and Fargate provisions and runs it automatically. Fargate + ALB is the standard serverless container architecture pattern.
AWS Step Functions — The Workflow Orchestrator
Use Step Functions to automate multi-step business processes. For example, "user registration → email verification → welcome message → CRM enrollment" — a sequential workflow defined visually.
Each step can be a Lambda function, SNS notification, SQS message, DynamoDB update, and more. Step Functions automatically handles data passing between steps, error handling, retries, and parallel execution.
Exam Key Points
"Asynchronous communication between components, acts as a buffer" — Amazon SQS
"One event delivered to multiple targets simultaneously" — Amazon SNS fan-out
"SNS publishing to multiple SQS queues at once" — SNS+SQS fan-out pattern (very frequently tested)
"Store and analyze failed messages separately" — Dead Letter Queue (DLQ)
"Guarantee exact message ordering and deduplication" — SQS FIFO queue
"Route events based on content, filter with rules" — Amazon EventBridge
"Build a serverless REST API" — API Gateway + Lambda
"Run containers without managing servers" — AWS Fargate (with ECS or EKS)
"Run multiple Lambdas in sequence or in parallel" — AWS Step Functions