2.2. Developing with Containers on AWS

💡 First Principle: Containers package your application with its dependencies into a portable unit — unlike serverless, you control the runtime environment but take on more operational responsibility.

Why use containers when Lambda exists? Because some workloads need long-running processes, custom runtimes, or more than 15 minutes of execution time. Without containers, developers must re-architect applications that don't fit the serverless model. Consider a Java application with a 30-second startup time — Lambda cold starts would destroy user experience, but ECS keeps the container warm and ready. The trade-off is clear: containers give you more control but require you to think about orchestration, scaling policies, and image management.

Think of Lambda as ordering a taxi (show up, ride, leave) versus containers as leasing a car (more flexibility, but you handle maintenance, parking, and fuel).