As more people and organizations migrate their activities and workloads to the cloud, we find it as important as ever to start out agreeing on definitions. Most people assume that they said exactly what they meant, that what they said is exactly what others heard, and even more insidiously, that their definition of terms is universally applied. When things go off-track at the very beginning it’s so much harder to end up at the intended destination.
We said in an earlier post that we define the cloud in a self-serving way as any infrastructure that operates outside your own facilities. It’s easier to use that generalization to get the conversation going. But once it’s going, there is additional value to appreciate the distinctions from more precise definition of the services available.
- Colocation - The resources provided include the building, power, cooling, access to bandwidth, and physical security. The customer is responsible for servers, storage, and software. Space in the facility is often leased by the rack, cabinet, cage, or room, though billing models vary.
- Private Cloud - The resources are dedicated to a single customer. They may exist on or off premises, and be owned, managed, and/or operated by the customer, a cloud service provider, or a combination.
- Community Cloud - The resources are available to a community of organizations with common characteristics (e.g., mission, security requirements, policy, application category, or compliance considerations). They may exist on or off premises and be owned, managed and/or operated by any or all of the users, a CSP or a combination.
- Public cloud - The resources are open to all customers through a cloud service provider. They exist on the premises of the CSP. They may be owned, managed and/or operated by a business, academic or government institution, or any combination.
Public cloud is the multi-tenant model that most people implicitly think of when they say ‘cloud’. To level set the standard against which all services are judged, a cloud architecture must have these attributes:
- On-demand self-service - The customer is able to self-provision computing resources, (e.g., server, networking, storage) without requiring support from the provider
- Broad network access - The customer is able to access the computing resources over the network through the use of desktops, laptops, mobile devices, servers, etc., and via a wide number of data transport providers
- Resource pooling - Computing resources are pooled to serve one or more customers
- Rapid elasticity - Computing resources are rapidly provisioned and released to match the demand
- Measured service - Use of computing resources is monitored, controlled, and reported. A desired model for many companies is to take advantage of the measured service as a utility model, where you pay for exactly and only what you consume.
As mentioned in our earlier post, most solutions now and into the foreseeable future will be a hybrid mix of components that live in the public cloud, in private segments of the cloud, and on company premises.
There is one final distinction we wish to use to define cloud: anything that is delivered as a Service (XaaS) is cloud based. The variety of infrastructure that is employed by the provider may not be important to the consumer of the service as long as it meet the performance characteristics expected from cloud service. With businesses becoming increasingly digital, there is some basis to state that every company is a cloud company. More on that in a subsequent blog post.