What is serverless computing (serverless processing) / FaaS?

Functions as a Service (FaaS)

FaaS is the basic building block of serverless architecture. It relies on running small pieces of code (functions) in response to specific events (events), such as an HTTP request (e.g., an API call), a file upload to an object store, a message in a queue, or a change in a database. The FaaS platform automatically scales the number of function instances according to the number of incoming events, ensuring high availability and cost efficiency. The developer provides only the function code, and the platform takes care of the rest.

Key features of serverless/FaaS architecture

• Server abstraction: No need to manage servers, operating systems, patching, etc.
• Auto-scaling: the platform automatically scales resources up and down (even to zero) to match the current load.
• Pay-per-execution: Paying only for the actual computing time and resources consumed during function execution, not for idle servers.
• Event-driven: Functions are typically triggered in response to events from various sources.

Backend as a Service (BaaS)

In addition to FaaS, serverless architecture often uses Backend as a Service (BaaS). These are off-the-shelf cloud services provided by vendors that implement typical backend functionality, such as user authentication (e.g. AWS Cognito, Firebase Authentication), database management (e.g. AWS DynamoDB, Firestore), file storage (e.g. AWS S3) or push notifications. Developers can use these services instead of building them themselves.

Benefits of a serverless approach

• Lower operating costs: Paying only for actual resource consumption can be much cheaper than keeping servers running all the time, especially with fluctuating loads.
• Faster development (time-to-market): Developers can focus on business logic without wasting time on infrastructure management.
• Automatic scalability and high availability: the platform takes care of scaling and fault tolerance by itself.
• Reduction in operational complexity: Fewer infrastructure management tasks.

Challenges and limitations of serverless

• Cold start (Cold start): The first call to a function after a period of inactivity may take longer because the platform must initialize its environment.
• Resource and execution time limits: FaaS features typically have limits on execution time, available memory or deployment package size.
• Complexity of distributed systems: Debugging and monitoring applications composed of multiple BaaS functions and services can be difficult.
• Vendor lock-in: Heavy reliance on specific services and mechanisms of a given cloud provider.
• State management: FaaS functions are typically stateless, requiring the use of external services (e.g., databases) to store state.

Examples of FaaS/Serverless platforms

The major cloud providers offer their serverless platforms: AWS Lambda, Azure Functions, Google Cloud Functions.

Summary

Serverless computing and FaaS represent the next step in the evolution of cloud computing, offering an even higher level of abstraction and cost efficiency. They allow the rapid building of scalable, event-driven applications without the need to manage servers. While they come with some challenges, they are an attractive option for many modern applications, especially microservices, data processing or backends for mobile and web applications.