Building Scalable APIs Using Serverless

### Overview Presented at the Open Source Summit organized by the Linux Foundation, speakers Qëndresa and Albiona discuss modern cloud-native architectures with an emphasis on building scalable APIs using serverless computing and Function-as-a-Service (FaaS). Drawing from their experiences running tech communities and development projects in Kosovo, they contrast traditional infrastructure and microservices with serverless workflows, walking through practical configuration, deployment, debugging, and trade-offs using the Serverless Framework, AWS Lambda, and Amazon DynamoDB.

### Key Topics - **Cloud Computing & Cloud-Native Foundations**: Shifting from owning servers and managing hardware to treating infrastructure as a utility. The talk examines containerization (packaging application code and dependencies via Docker) versus virtual machines, container orchestration, and breaking monolithic applications into modular microservices with defined APIs. - **Cloud Service Delivery Models**: Comparing Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Function as a Service (FaaS), highlighting differences in resource management, operational responsibilities, scalability, and security obligations. - **Serverless Paradigm & FaaS**: Understanding that serverless still involves servers, but execution, resource allocation, and scaling are managed entirely by cloud providers such as AWS Lambda, Azure Functions, or Google Cloud Functions. - **Implementation via Serverless Framework**: Structuring configuration using `serverless.yml`, modular YAML files for routes/functions, environment variable management via `.env`, and plugins such as `serverless-offline`. - **Deployment and Operational Monitoring**: Demonstrating a live deployment of CRUD API handlers with DynamoDB, alongside monitoring and observability using AWS CloudWatch and the Serverless Framework dashboard.

### Technical Details - **Architecture & Infrastructure Layers**: The presentation outlines how application development evolved past physical hardware, virtualization, and OS management into runtime and business logic abstractions. Unlike VMs that virtualize at the hardware layer with separate operating systems, containers share the OS kernel, isolating CPU, memory, storage, and networking resources. - **API Implementation Walkthrough**: A live project demo showcases an e-commerce API ('Uru Store') structured using AWS Lambda in `us-east-1` connected to Amazon DynamoDB table (`serverless-presentation-example`). The configuration defines HTTP events (e.g., POST and GET endpoints for `/jewelry`), IAM role statements for database access, execution timeouts, and handler paths. - **Debugging and Observability**: Troubleshooting serverless architectures often presents challenges due to cryptic execution errors. Recommended approaches include enabling `serverless debug`, reviewing centralized metrics and logs in AWS CloudWatch, and tracking request volumes and invocation failure rates using the Serverless Framework dashboard.

### Takeaways - **Developer Velocity**: Serverless models allow development teams to focus primarily on writing business logic without provisioning or manually maintaining servers, while benefiting from built-in pay-per-use scaling. - **Trade-offs & Challenges**: Adopting serverless introduces risks including vendor lock-in, limited system-level control, porting difficulties between providers, complex testing and limited local debugging options, and a necessary mindset shift regarding architecture. - **Shared Security Model**: Under serverless and PaaS models, underlying platform and hardware security relies on the cloud provider, but consumers remain responsible for secure configuration, IAM roles, application logic, and credential management.