First the physical infrastructure moved to cloud, well yeah not literally cloud but to some location where it could act as 'Cloud'. As if for the uninitiated, this was not enough, we now have Serverless computing. The servers are always needed for computing so that's not going to change (atleast for sometime until next disruption happens).
So what exactly is the Serverless computing? Wikipedia has rightly named 'Serverless computing' as a misnomer. In its simplest form, it is a shift from managed services to managed tasks. With Serverless computing, we do not need to allocate the actual servers or machines. We simply define the functions or tasks that we need to accomplish.
A quick example to understand this is a task of resizing the images as and when the requests comes from another application. Now in conventional sense, one would need to design an application to communicate, store, resize, process, share the images. This application would need to run on a server which would need both money and resources even when they are not needed. With Serverless computing, we can write a function to pick up image from a location, resize and then store them. These will be the only tasks performed by this function and in result the money would also be charged based on each function call.
In its simplest form, its a Function-as-a-service code execution platform.
Two leading cloud computing companies have this kind of functionality. AWS calls it 'Lambda' and Microsoft Azure calls it 'Functions'. AWS Lambda and Microsoft support languages like Node.js, Python, and C#. However, AWS Lambda also provides support for Python and Java, while Azure Functions provides support for F# and PHP.
These services provide immense capabilities and scope for efficient operations to anyone looking for scalable and cost effective approach. Not everything can be done using Serverless computing but majority of the standalone applications can use benefits of Serverless computing.
So the Serverless computing is not so Serverless!