Cloud computing is the delivery of computing resources — processing, storage, databases, software — as a service over the internet, drawn on demand from a shared pool. The user does not own or maintain the hardware; a provider runs large data centres and the user pays for what is consumed, scaling capacity up or down as need changes.

Two ideas in the correct option carry the definition. Shared refers to the pooled infrastructure that many customers draw from, and on demand refers to self-service provisioning, where resources are obtained in moments without a purchase or an installation.

The familiar service models follow from this: IaaS supplies raw servers and storage, PaaS supplies a development platform, and SaaS supplies finished applications such as web-based mail or document editors.

Cloud computing is not a unit of measurement — computing power is measured in quantities such as FLOPS or clock speed, and the cloud is a delivery model, not a metric.

Nor is it confined to high-end weather forecasting. Such work runs on supercomputers built for tightly coupled parallel processing, which is a different arrangement from a pool of shared commodity servers.

It is also not a video transfer protocol; file and media transfer over the web is handled by protocols such as HTTP, FTP and RTMP.

Hence cloud computing is internet based computing which provides shared computer processing and data to digital devices on demand.