Day 14: AWS Compute — Managed & Serverless

On Day 14, we delve into AWS Compute services that offer more abstraction and require less manual management from users. Day 13 introduced EC2, ELB, and Auto Scaling, which demand full control over infrastructure setup and management. Today, we explore services where AWS assumes more operational responsibilities.

AWS Elastic Beanstalk serves as a Platform-as-a-Service (PaaS) option built upon EC2 fundamentals. With Beanstalk, developers upload application code and configuration files. Beanstalk then provisions EC2 instances, a load balancer, and an Auto Scaling Group on your behalf. Think of it as dining at a restaurant rather than cooking at home: AWS manages the underlying servers, letting you focus on application deployment.

Users can still access and modify the underlying EC2 instances should they choose, offering a balance between simplicity and control. Beanstalk is ideal for teams seeking rapid application deployment without setting up their CI/CD pipeline from scratch. However, as needs grow, scaling from Beanstalk to raw EC2 might become necessary for more granular infrastructure control.

Amazon Lightsail presents a different abstraction level. It functions as a virtual private server (VPS) service, allowing users to spin up pre-configured instances tailored for common applications like WordPress, GitLab, Node.js, and more. Lightsail instances come with predefined stacks, simplifying the setup process. Unlike EC2's pay-as-you-go model, Lightsail features fixed pricing, making it more predictable for small businesses and non-technical users.

Its limitations include no support for Auto Scaling and custom networking configurations. When faced with scaling requirements or custom networking needs beyond its capabilities, transitioning to EC2 + ELB + Auto Scaling becomes the logical next step.

AWS Lambda represents the pinnacle of serverless computing, eliminating the need for server management altogether. Developers write code as discrete functions, which Lambda executes in response to specific triggers—such as HTTP requests or scheduled events—without provisioning underlying servers. Pricing is based on actual execution time and the number of requests, not idle capacity, making it cost-effective for short, event-driven tasks.

Key considerations include the 15-minute execution limit, which is suitable for lightweight tasks such as image resizing or lightweight API backends. Lambda integrates seamlessly with AWS services like API Gateway and S3, automating processes like scheduled EC2 instance start/stop cycles and enabling event-driven automation through services like EventBridge.

EventBridge, another serverless offering, acts as an event bus for routing and processing events across AWS and SaaS applications. Unlike SNS, which focuses on pub/sub messaging for notifications, EventBridge specializes in defining and triggering actions based on event patterns. This makes it particularly useful for orchestrating workflows, such as stopping EC2 instances at specific times by invoking a Lambda function.

While both EventBridge and SNS deal with event-driven automation, SNS is better suited for simple notification distributions.

In summary, these AWS compute services offer varying degrees of abstraction and control, catering to diverse application needs—from fast, easy deployments with Elastic Beanstalk and Lightsail to serverless automations with Lambda and EventBridge. Choosing the right service depends on factors like scalability requirements, technical expertise, and specific application demands. For further exploration, visit the Hashnode series on AWS Compute services.