A public cloud can provide access to computing resources that many companies would otherwise never be able to afford. The arguments for the cloud are well-known by now, but they remain compelling—no up-front costs, virtually unlimited computing power on demand, and highly efficient pricing where customers pay only for resources used. There’s also less pressure on corporate IT departments that are charged with managing the infrastructure and budgeting for new equipment to keep up with demand.
But concerns about security and loss of control in public clouds have led to an alternative model—the internal cloud—that replicates the cloud environment inside the corporate firewall. Within these boundaries, enterprise users can provision computing resources as needed, using the cloud’s self-service capabilities while leveraging data center services. Internal clouds are often referred to as private clouds, but since private clouds can also be found in external environments, the “internal” designation is a more precise term for what we’re talking about here.
With servers, applications and data within the enterprise walls, internal clouds can provide many of the benefits of cloud computing without the potential risks when the computing environment is provided by a third party. Unfortunately, the economics of internal clouds makes them inherently less efficient than the public cloud, especially as new technology makes the public cloud safer and more reliable.
Here are some of the reasons:
- Infrastructure costs: Deploying an internal cloud requires building out the infrastructure to support the needs of all enterprise users. In addition to acquiring the necessary hardware and software, this includes things like configuring the network, allocating storage, paying the electric bill, and providing square footage for the equipment. Plus, all of this infrastructure has to be managed and supported on an ongoing basis.
- Over-provisioning: Just as with traditional IT infrastructure, anticipating resource requirements for an internal cloud is difficult since applications run at varying usage levels. Some applications consume resources fairly steadily while others require occasional bursts of massive computing power. Companies may have no choice but to over-provision, where some equipment sits idle most of the time in order to have resources available for peak periods. (Meeting these short-term usage needs is one example where the elasticity of public clouds really pays off.)
- Building the management plane: Building an internal cloud is about more than just virtualization. One of the key benefits of cloud computing is the self-service aspect, where users can access resources as needed via a self-service portal, which then executes in the cloud automatically without administrator intervention. Building this control plane is another substantial step when implementing an internal cloud.
For all of these reasons, the internal cloud carries a cost overhead compared with leveraging the resources and self-service framework already provided by public clouds. Also, much of the same protection and control offered by internal clouds is becoming increasingly available in an external public environment. So enterprises may find that a public cloud may provide a more cost-effective home for many (or even most) of their applications.
Internal clouds can make sense for the most business-critical requirements or those that need specialized hardware or have regulatory compliance issues. Focusing internal resources on these types of applications would greatly reduce the investment required to build, sustain and grow enterprise infrastructures. While some applications may always need to run internally, many companies have dozens of others that could run more cost-effectively today in a public cloud, providing computing power on demand along with required security, agility and scalability.