What is Cloud Computing?
Cloud computing is on-demand access to computing resources — servers, storage, databases and higher-level services — operated by a provider and consumed over a network.
Hover a route to see which characteristic it demonstrates; click for a short explanation.
What you are looking at
On the left is the customer: a person, a team or an application that needs computing capacity.
In the middle is the network. Every cloud interaction is a request travelling over a network to the provider.
On the right is the provider's environment, where capacity is offered as services: compute, storage, databases and higher-level managed services.
At the bottom sits the physical infrastructure. It still exists — the provider owns and operates it, and shares its capacity between many customers.
How it works
- 01Request. You ask for a resource through a portal, CLI, template or API — no purchase order, no hardware delivery.
- 02Provision. The provider allocates capacity from a shared pool, usually in seconds or minutes.
- 03Consume. Your users and applications reach the resource over the network.
- 04Scale. You add or remove capacity as demand changes, manually or automatically.
- 05Pay. Usage is metered and billed for the period the resource was consumed.
Why it matters
- It changes computing from a capital purchase into a consumable service, so an idea can be tested without buying a server.
- Elasticity means capacity can follow demand rather than being sized for the worst day of the year.
- Responsibility is shared: the provider operates the infrastructure, while the customer remains responsible for configuration, data and access.
Common misconception
“"The cloud means there are no servers any more."”
The servers are still there. They live in the provider's data centres, and the provider operates them for many customers at once — you simply consume their capacity as a configurable service.
Real-world example
A small online shop expects normal traffic all year and ten times that on launch day. Instead of buying servers for the peak, it runs on managed compute plus a managed database, scales out automatically for the launch, scales back down the next week, and pays for the extra capacity only for the days it existed.