What is serverless computing?

The official literature of Amazon Web Services (AWS), one of the de facto serverless providers, defines serverless computing as follows:

Serverless Computing allows you to build and run applications and services without thinking about servers. Serverless applications don't require you to provision, scale, and manage any servers. You can build them for nearly any type of application or backend service, and everything required to run and scale your application with high availability is handled for you.

It's worth exploring the implications of this definition as our first step into the serverless world.

....build and run applications and services without thinking about servers.

Producing software involves much more than just writing code. The code that the development team writes exists to solve a real-world problem, and needs to be available to the intended audience. For your code to serve the world, it (traditionally) has to exist on a server. The server itself has to be created (provisioned) and made capable of handling the workload that the business demands. The capabilities of a server are defined in many ways, like its processing power, memory capacity, and network throughput, just to name a few. These parameters are so vast and deep that they have spun up a vast market of jobs that businesses require. The jobs go by titles such as infrastructure management associate, operations associate, and, more recently, DevOps engineer.

It's the responsibility of these folks to evaluate and manage the hardware properties. That is what the definition highlights when it states, thinking about servers.

...Serverless applications don't require you to provision, scale, and manage any servers.

Serverless computing takes away the aforementioned need to think about the servers and other hardware resources.

...nearly any type of application or backend service.

As a paradigm, serverless computing can be applied to any solution that requires a backend or a piece of architecture and code that is not (or cannot) be exposed to the general public (loosely termed clients).

...everything required to run and scale your application with high availability is handled for you.

In the serverless paradigm, there are computational hardware assets, like servers, the management of these computational assets is not the developer's concern.This turnkey management is offered on a pay-as-you-use models keeping the costs as high or as low as the utilisation of the assets necessitate.

So, serverless computing itself is a misleading term, or misnomer. There are computational hardware assets serving your code, but their management is the cloud providers' problem.

This frees the companies adopting this paradigm from the overhead of the mundane, but equally important, tasks of tending and managing systems that behave well in production. It allows them to have a laser-sharp focus on their most valuable task - that is, writing code.