Serverless

What it means

Serverless is a cloud model in which developers upload code and the provider runs it on demand when a request arrives, charging only for actual execution time. The company manages no servers, updates or capacity; servers of course exist, but the provider looks after them.

A typical example is functions triggered by events: a form submission, a file upload, an API call or a schedule. With zero traffic almost nothing is paid, and when traffic grows the provider adds capacity automatically.

The model differs both from renting servers and from containerisation, where the company still decides how many instances of what size run. Serverless shifts more responsibility to the provider and fits well with a DevOps approach with automated deployment.

How it works in practice

For smaller websites and applications with uneven traffic, serverless can be cheaper and simpler than an always-on server. A static site with a few functions, for example to process a contact form, can be run at very low cost.

Design must account for platform limits: maximum function run time, memory limits, cold-start latency and how data are stored. Secret keys, such as those for an email service or form protection, are stored as platform environment variables, not in the code.

Costs must be monitored. Although billing is per use, an unexpected surge in requests, for example from an attack or a bug, can drive up the bill significantly. Budget alerts and limits help.

What to watch out for

Vendor lock-in is a real risk. Code written for a specific platform is hard to move if prices or terms change. It is wise to keep business logic separate from platform specifics.

Debugging and monitoring distributed functions is harder than with a single application on a server. Without good logging, errors such as a silent failure in sending form submissions may go unnoticed for a long time.