Cloud Deployment Models

Decorative image showing Private cloud, hybrid cloud, and public cloud symbols.
Private-Hybrid-Public Cloud Symbols

BeeBright/Essentials Collection/iStock

Cloud computing refers to the use of remote servers over the internet (instead of via on-premise servers or devices) for the purpose of sharing resources.

There are three categories of cloud deployment models: public clouds, private clouds, and hybrid clouds. The models provide similar benefits, including cost-effectiveness, performance, reliability, and scale, but differ in their security and management demands. The most suitable deployment method depends on your business needs and circumstances.

Public Cloud

The most popular deployment model is public cloud. In this model, the physical cloud infrastructure is located on the cloud provider's premises, and it is shared by all cloud customers. This means that a virtual server can run on a physical host on which virtual servers of other customers are running. Depending on the loads placed on other customers' virtual servers, the host may allocate fewer resources to the virtual server. Although there is a mechanism in place ensuring that no server will be "starved to death," the fluctuations in CPU or memory allocations can be recognizable.

Private Cloud

For that reason, organizations may want to get exclusive access to physical resources implementing the cloud. This is called private cloud. It can happen on- or off-premises of the customer.

When customers provision a virtual server on the premises of a cloud vendor, they can get a dedicated physical host and can request that subsequent virtual servers of that customer will be provisioned on the same host. Some vendors call it a dedicated cloud model.

Because of security or control considerations, some organizations insist on having the physical cloud infrastructure located on their premises. This is the classical private cloud deployment model, and some vendors call it local cloud. The customer can take advantage of cloud elasticity and adaptability, but the customer retains a higher degree of control over it.

In both cases, the virtualization and cloud control software can be owned and maintained by the organization itself, or it can be provided by a third-party vendor. Customers can provision dedicated servers in an off-premises cloud and have the cloud provider create and manage virtual dedicated servers, or they can install virtualization software in them (for example, open-source OpenStack, or VMWare vCenter), and manage their cloud themselves. Similarly, using on-premises hardware, organizations can outsource cloud creation and management to a third-party vendor, or do it themselves.

Hybrid Cloud

hybrid cloud can be a mix of any of the above models. This may be a powerful way of using the cloud paradigm, allowing organizations to run legacy or critical applications on premises, and applications better suited for cloud, like backup or new "born for the cloud" software off premises. Hybrid cloud requires a more advanced network setup to ensure safe and efficient on-off premises communications, but many organizations are using it.

It is worth noting that any of the deployment models above can support any of the service models, including IaaS. That means organizations can run clouds providing pure infrastructure services on and off premises, shared or dedicated.

References

Anderson, E. (2014, July 14). How to explain private, public, and hybrid cloud to your spouse [Blog post]. Retrieved from https://www.ibm.com/blogs/cloud-computing/2014/07/explain-private-public-hybrid-cloud-spouse/

IBM. (n.d.). IaaS, PaaS, and SaaS—IBM cloud service models. Retrieved from https://www.ibm.com/cloud-computing/learn-more/iaas-paas-saas/