Private Cloud Infrastructure Design: Go Beyond Best Practices

Of all of the possible benefits of a private cloud infrastructure, one of the most valuable is flexibility. With a properly designed private cloud infrastructure, the data center environment can fluidly shift with the business. This allows new applications to be deployed to meet business demands as they're identified, and legacy applications to be removed when the value is no longer recognized.

In order to have an environment capable of this rapid application deployment cycle, an infrastructure must be in place that can handle it. Hardware and software must be properly architected to provide both application and hardware scalability. This requires a rethinking of the design principles that have brought IT to where it is now.

The first step is dismissing the term, and more importantly the notion of best practices, as there is no such thing. Rather than look for "best practices," search for "current recommended practices" that apply to your business and intended deployment. The entire concept of best practices is a crutch at best, and typically does more harm than good. Every environment is different, and technology changes daily. By the time a practice becomes "best," the industry has moved on. I'm fond of saying, "Today's best practice is tomorrow's bad idea."

The next step is to arm yourself with the most powerful question in data center design: "Why?" Apply this to every infrastructure decision: Why should you build a 3-Tier network? Why not use DAS? Why virtualize individual servers? Why build a flat mesh network? Why use VLANs for segregation? Why not use more Layer 3? Why use blades? Why use forced air cooling? Why not use blades? Why use DAS? And many other whys.

The traditional silo and application-specific hardware architectures will not suffice for cloud scale and rapid flexibility. In order to produce flexible pools of resources to support a changing environment of applications and services, the environment must be designed from the ground up with that in mind. Compute, network and storage must be designed with scale and performance first and foremost in mind.

This rethinking should factor into all aspect of the design. For example, ask yourself why you typically spend significant additional dollars for redundant hardware components on servers, then ask yourself if this is still applicable in a private cloud model. If the upper-layer software, virtualization, applications, etc., handle failover fluidly at a higher level, does the server require costly hardware redundancy?

In many cases, pre-architected private cloud hardware stacks can be used to build an infrastructure foundation. With these in place, you have added vendor support and a tightly integrated hardware stack. From there, you'll be able to build a software environment with the required tools to deploy the applications that run your business. Again, legacy decisions and practices will lead to the legacy infrastructure you're trying to break from. Ensure that your cloud management software can handle your expected scale.

Disclaimer: This post is not intended as an endorsement for any vendors or products mentioned.