A cloud is a distributed set of resources that are individually known as cloud pieces but collectively known as digital infrastructure.

When a cloud's resources are all under the ultimate authority of one entity, this is a fully centralised cloud. When each resource has its own distinct ultimate authority, this is a fully de-centralised cloud. Anything in between is typically referred to as a (partially) de-centralised cloud. The level of de-centralisation can be measured through numerous methods (e.g. nakamoto co-efficient) to estimate the minimal number of distinct authorities required to collaborate to compromise the entire cloud.

These concepts apply equally and independently at each level of the software-hardware stack. For instance, a cloud can be de-centralised at the hardware level but centralised at one or more of the software levels. And vice-versa.

The mega clouds typically have a huge number of cloud resources at their disposal. They are capable of handling large enterprise requirements.

The mini clouds typically have a small number of cloud resources at their disposal. They are designed for personal use or small enterprises.

The challenger clouds typically have a large number of cloud resources at their disposal but are still limited in their capability to handle the requirements of large enterprises. They are typically suited to SMEs or large enterprises utilising multiple cloud providers (i.e. cloud aggregation).

A homogenous cloud typically focuses on providing one type of cloud resource. This is prevalent within de-centralised clouds although there are some examples of mini centralised clouds that also operate in a homogenous manner.

Conversely, a heterogenous cloud aims to provide many different types of cloud resources. In the case of mega centralised clouds, they will typically aim to provide a comprehensive set of cloud resources to handle all user requirements.

A public cloud is typically open for use by anyone subject to laws and regulations outside of the control of the cloud provider.

A private cloud has a restricted set of users. Typically, a private cloud will belong to an enterprise and will be used solely by that enterprise. However, a cloud is still private if external access is given but is restricted to a small number of permissioned users.

A cloud aggregation strategy is when multiple clouds are utilised to satisfy the requirements of a user. The collective term for the clouds is known as a multi cloud.

A multi cloud that incorporates both a private cloud and a public cloud (i.e. at least 1 of each) is known as a hybrid cloud. Sometimes this is a temporary state that occurs during a migration from one to the other. In other cases, it is by design.

A multi cloud that seeks to abstract away the component clouds that make it up, is typically referred to as a cross cloud.

A multi cloud whose component clouds are from a restricted set of curated and trusted providers is a closed multi cloud. Whereas a multi cloud whose component clouds are from an unrestricted set of providers (as far as laws and regulations allow), in which any cloud provider is free to make their cloud's resources available to users, is an open multi cloud.

A cloud architecture is the set of cloud pieces (i.e. cloud resources) that come together to produce a cloud application that serves a purpose.

A dynamic cloud architecture is one whose component cloud pieces can change on-the-fly in response to the needs of the cloud application with regards to its purpose. For instance, improving application responsiveness, handling increasing data storage requirements, managing a sudden high load etc.

A fluid architecture is one whose component cloud pieces can change on-the-fly in response to external factors unrelated to the performance of the cloud application. For instance, laws and regulations, company policies, cost-optimisations etc.

A zero trust architecture is one in which every cloud piece within the architecture applies IAM/PAM principles whenever and however it is accessed and from whereever it is accessed. It takes the view that a breach of the architecture is always ongoing and seeks to limit the potential of the breach at all times.