Google Cloud Functions

Many web applications consist of an application runtime, an HTTP server technology, and a series of externally invokable operations that perform some unit of work. Taken to an extreme, service abstraction can extend to the application layer itself, removing the need to manage the application's runtime. This level of abstraction is known as FaaS, where invokable operations are the base unit to be managed.

FaaS is a relatively new and exciting development in the public cloud space, and Google Cloud looks to fill this need with Cloud Functions. Cloud Functions allows users to focus on building atomic units of operation in the form of a collection of inputs, logic to act on the inputs, and an output. These functions can then be deployed to GCP where they can be invoked as needed. Invocation can occur from a number of events and triggers such as HTTP requests, changes to Cloud Storage objects, and Pub/Sub messages. This makes Cloud Functions an ideal candidate for event-driven architecture.

Scaling and availability are extremely simple with FaaS, and the on-demand model and lack of application runtime allows FaaS solutions to dramatically undercut the cost of other compute options. FaaS is often referred to as serverless technology, though that term is increasingly used when referring to PaaS solutions as well.