"Public clouds are homogenous; private clouds are heterogeneous." That's the statement Bob Muglia, former president of Microsoft's Server and Tools Business, made at a developer conference a few years ago. And this statement was flawed on two points.
The relative diversity of cloud computing resources has never been part of the cloud discussion, nor is it a differentiator among cloud deployment models. The complexities of implementing the on-demand elasticity and resource-pooling characteristics of a cloud also make heterogeneity impractical. However, Muglia's quote was useful to highlight the perception that there are technical differences between public and private clouds. In reality, cloud deployment models only differentiate the degree of infrastructure sharing that's permitted in an environment. Aside from that, private cloud is no different from any other cloud. And cloud washing only further confuses matters.
The National Institute of Standards and Technology definition of cloud shows three sections: essential characteristics, service models and deployment models. Cloud washing is tough to define since the three sections have no interdependencies; there is no structure that says public clouds have one set of characteristics and private clouds have a different set. Selecting a deployment model is primarily a business decision with a small security component. Do the requirements of the organization permit the sharing of infrastructure? If not, a private cloud is the correct path.
Where does cloud washing come into play? Private clouds are viewed as a combination of cloud's hyperefficiency and ease of provisioning with the control that enterprise IT wants. This permits participation in the cloud while lowering the perceived risk. However, the potential return on investment is low, as it offers only a slight improvement over current methods like virtualization.
The Capex vs. Opex cloud promises
Much of the cloud conversation has been focused on public clouds, and the public cloud's many benefits only apply at a much lesser degree than those of private clouds. A big topic with public cloud has been the exchange of the Capex model for the Opex model, which moves spending from a large up-front number to one distributed over time.
There are several problems with this model in the enterprise, foremost of which is the challenge of variable cost in a budgeting system based on fixed costs. But assuming the variable cost model can work in the enterprise, on-premises private clouds do not offer this benefit.
Capex is still required ahead of demand. Most likely, it will involve a larger spend than with a legacy model.
Private cloud essentials
Now that we've covered the difference between the benefits of public cloud versus private cloud, we should look at some private cloud characteristics. Here's a brief analysis of the five crucial cloud features, as applied to the private cloud model:
- On-demand self-service: good, but it creates a risk of resources being allocated and parked.
- Broad network access: good, but storage strategy needs careful consideration.
- Resource pooling: good, but homogeneity is needed if self-service requirement is to be met.
- Rapid elasticity: good if enough capacity is available. But enough capacity requires a reserve; a reserve requires Capex beyond the current needs of the organization. Utilization requires an application architectural model that uses the cloud-scaling model.
- Measured service: Getting better, but it's only useful if accounting is integrated.
With the current focus on self-service, private cloud offers little improvement over the existing virtualization model and maintains the same risks as the existing model -- complexity, poor resource utilization, etc. It's important to establish a process to ensure thoughtful allocation of resources and the recovery of unused resources.
Private clouds can meet all of the requirements and provide the value of cloud computing; however, additional effort is required to implement and apply them. Remember, clouds solve Web-scale problems, and those methods can help fix some enterprise-scale problems.
On-demand self-service and rapid elasticity are the two key characteristics of cloud computing, but leveraging them in the enterprise requires a very different approach to consuming IT resources. For example, workloads that are separated by a schedule can use the same resources. This requires that each workload can consume a common set of resources rather than custom-configured servers. This affects application development and operation.
Seasonal workloads are best run in a hybrid cloud model, which can significantly affect development and operation. Creating the core of the application to run in a private cloud during off-peak times and drawing in resources from an external cloud for peak periods requires a sophisticated architecture.
So, is a private cloud for you? To determine if it's the best cloud model for your organization, ask these four questions:
- Do you have a business problem that can be solved using cloud techniques?
- Are your security and regulatory challenges sufficient to block consideration of public cloud?
- Is the up-front Capex of a private cloud offset by improved productivity and/or profits?
- Can your organization change how it applies technology to business solutions?
If a private cloud best serves your business needs, then deploy one. But make sure it's a real private cloud -- not a cloud-washed virtualized data center.
About the author:
Mark Eisenberg has been doing mobile app development since 2005. He joined the nascent Windows Azure sales team at Microsoft despite being an early cloud skeptic. Now, after embracing the cloud and its technological potential, he combines cloud and mobile expertise with his technological background to help clients realize real value from their technology investments. Mark is also a seasoned business development professional with more than 20 years of experience. He started his career in software development and has since maintained his technology edge, most recently adding cross-platform mobile development skills. His sales career began when he joined Intel's channel and has included positions at other communications-focused firms prior to Microsoft.
This was first published in January 2014