What Serverless Really Means

Marcus's startup ran a daily batch job that scraped competitor pricing — about 12 minutes of compute, once a day. They paid $34/month for a t3.small EC2 instance to host it. After they rewrote it as a Lambda function, the next bill was $0.04. Same workload, 850x cheaper.

But that's not why this course matters. Serverless is not really about cost. It's about everything you stop owning.

What "Serverless" Actually Means

The word is misleading — there are still servers, AWS just runs them. The useful definition has four parts:

• No servers to provision — you do not pick instance types, you do not size a fleet;
• Automatic scaling — from zero to thousands of concurrent executions, with no config;
• Pay-per-use — you are charged for invocations and compute time, not idle capacity;
• Built-in availability — the service is multi-AZ by default. No "high availability" project. A service is serverless when it gives you all four. Lambda, API Gateway, DynamoDB, S3, SQS, SNS, EventBridge, Step Functions — all serverless. EC2, RDS, ElastiCache — not. Aurora Serverless and Fargate sit between, depending on how strictly you define the term.

Why This Matters for Developers

The shift changes what you spend your time on:

• You stop patching operating systems;
• You stop sizing instances and forecasting capacity;
• You stop building "highly available" deployments — they are highly available by default;
• You start composing services like Lego bricks. A working serverless app on AWS is often a Lambda function, an API Gateway in front of it, a DynamoDB table behind it, and an EventBridge rule on top of it. Four services, no servers, no Terraform for VPCs.

What This Course Covers

Across 11 chapters, you will cover the modern serverless and event-driven AWS stack:

• AWS Lambda — the execution engine for serverless compute;
• Amazon API Gateway — the front door, in two flavors (REST APIs and HTTP APIs);
• Amazon SQS — queues for decoupling work between services;
• Amazon SNS — pub/sub for fan-out to many subscribers;
• Amazon EventBridge — the event bus for content-based routing;
• Amazon Kinesis Data Streams — real-time streaming for high-volume event data;
• AWS Step Functions — orchestrating multi-step workflows across all of the above.

A Note on Cost

Serverless changes the cost model — but it does not always lower the bill. A constantly-running Lambda doing 10 million invocations per minute is more expensive than the equivalent EC2 fleet. The economics flip when:

• Traffic is bursty or irregular;
• Workloads have long idle periods;
• The team cannot afford 24/7 operations engineering. Marcus's batch job hit all three. Your workload might or might not. We will cover the math along the way.

A Note on the Exam

This is the densest course for DVA-C02. Roughly 30% of the exam is on Lambda, API Gateway, and the integration services in this course. Pay extra attention to:

• Lambda execution context, cold starts, concurrency limits;
• API Gateway integration types and authorizer flow;
• SQS visibility timeout and dead-letter queue patterns;
• Choosing between SNS, SQS, EventBridge, and Kinesis for a given use case. Let's start with the engine — Lambda.