What does serverless mean?

Serverless: definition and meaning

The name is misleading: of course, serverless applications also run on servers. What it means is that the company no longer has to take care of those servers. Provisioning, operating system updates, capacity planning and scaling are entirely the cloud provider’s job; operations become a service.

The best-known form is Function as a Service (FaaS): small pieces of code, called functions, start only when an event occurs—such as a file upload, a form submission or a call through an interface. Added to this are managed services such as databases, queues and storage, which are also used without any server maintenance of your own.

How pay-per-use billing works

A function is started only when it is needed and ends once the work is done. When requests surge, the provider multiplies the instances automatically; when there are none, hardly any costs arise. Billing is usually based on the number of calls and the compute time used, including memory.

There are also quirks: after a long pause, the first call takes a little longer (a cold start), a function’s runtime is limited, and it should not store any state of its own. Data belongs in a database or storage service. This requires a different architecture than a traditional server.

Real-world example: product images in an online store

A specialist retailer regularly uploads new product photos from suppliers to its store. Every image has to be converted into several sizes and modern formats. With serverless, a function starts automatically on upload, processes the image and saves it to storage—without a server idling overnight.

During a promotion, when many images suddenly arrive at once, several functions run in parallel. The retailer pays only for the images actually processed and has to neither reserve capacity nor monitor servers. For the permanently running store itself, however, this model would be less suitable.

Benefits, limits and imple­menta­tion

The benefits are low operating effort, automatic scaling and cost control under fluctuating load. The limits: strong dependence on the provider (vendor lock-in), more complex testing and debugging, runtime limits and hard-to-predict costs if a function is called en masse by errors or attacks. Under constant high load, reserved servers are often cheaper.

Nano Concepts works with you to determine which parts of your application suit serverless and implements event-driven functions, interfaces and automations—often combined with traditional servers or containers, so that every task runs on the technology that fits it best and costs stay predictable.

• Well suited: file processing, notifications, webhooks, scheduled tasks
• Less suited: constant load, long-running jobs, very strict latency requirements
• Set budget limits and alerts to avoid cost surprises
• Keep functions small and move data to a database or storage
• Assess provider dependence and exit options early