Serverless computing is a cloud model in which the provider takes over the operational work of running servers. In the 2019 paper “Cloud Programming Simplified: A Berkeley View on Serverless Computing,” Eric Jonas and co-authors define serverless computing as a system that “handles virtually all the system administration operations needed to make it easier for programmers to use the cloud.” The name is a slight misnomer: there are still servers, but the developer never provisions, patches, or scales them.
The Berkeley authors argue that serverless represents an evolution comparable to the move from assembly language to high-level programming languages: it raises the level of abstraction so that programmers reason about their application logic rather than the machines underneath it. Two properties distinguish it from earlier cloud models: the platform scales resources automatically with demand, and customers are billed only for the resources their code actually consumes.
The AWS Lambda product page illustrates the developer experience: run code “without thinking about servers or clusters,” with the provider handling all infrastructure management, applications that “scale seamlessly to meet unpredictable demands,” and “pay-per-use billing by the millisecond.” When there is no work to do, there is nothing to run and nothing to pay for.
In practice, serverless is usually delivered as two complementary services: Function-as-a-Service for running custom code, and Backend-as-a-Service for ready-made building blocks like databases, authentication, and storage. Together they let a developer assemble and ship an application without operating a server tier of their own.