In the evolving hybrid infrastructure philosophy of the future, it’s a virtual certainty that many designs will include a private cloud in the mix of cloud and service types. Quite a few organizations start their cloud journey with private cloud creation. Reasons for doing so vary: a learning exercise, executive pressure, shadow IT, a catalyst preparing IT and business operations for the future, hesitancy to relinquish control to third-parties, or perhaps it’s simply time to do so.
Many of these private cloud experiences will also be failures, not because it was a bad idea but because of missteps from the outset. Even the best of intensions can be undermined by ill-advised approaches organizationally, motivationally or technologically (or some combination). The result will be too expensive or complex, not meet enterprise requirements, ignore interoperability, and so on. There is a lot at stake. Getting it right by getting off to a positive start will help immediately as well as smooth the way for future requirements and rewards. Here are a few situations to consider.
IT Takes a Village
You can’t know what you’re building unless you know how it will be used, by whom and for what purpose. Doing what you’ve been doing, only better is misplaced optimism. Building services for the wrong workloads is not a road to success.
It stands to reason that if a private cloud puts more control into the hands of individuals, more flexibility into the organization and more speed into results, then active involvement from many parties builds shared accountability and purpose. Setting expectations for how much can and cannot be done early on helps frame priorities and surfaces unmet user needs. What constitutes success? Establish measureable, meaningful metrics that focus on organizational value, not operational changes.
Square Pegs in Round Holes
Traditional IT is the rhinoceros and private cloud is the oxpecker bird. There’s a symbiotic relationship between the two, but each has its own purpose. The rhino can run 30 miles per hour for short distances given it 7,500 pound frame, while the oxpecker … well, it’s a bird. Unless the operational model of your existing IT infrastructure is very mature and highly tuned, it’s best to not force fit your private cloud into it; they’re different animals.
Incubate your private cloud initiative, using separate teams and processes in the near term. Do so with a clear path to reintegration in the long term when synergy can improve efficiency without reducing your private cloud efficacy. Temporarily assign individuals to the incubation team to extend appreciation for private cloud differences and benefits, and to get them thinking about the future. Individuals’ roles and responsibilities will surely change with the expanding cloud footprint. As it does, a good number of IT and infrastructure & operations staff will become square pegs that may no longer fit.
Bite Off Only What You Can Chew
Some private cloud initiatives originate with a perceived “me vs. them” threat from outside cloud providers. The resulting private cloud strategy — and overall cloud strategy — is intended to keep the enterprise from using these external cloud providers, and continuing to invest in internal I&O. Attempting to match cloud service providers product for product, benefit for benefit in order to stave off competition will fail in the face of economies of scale and sustainable performance.
Instead, view cloud service providers as an extension to your own data center, not a replacement for it. Get smart about being a services broker. Use provider services as benchmarks. When they do not meet enterprise requirements for cost, service levels, security, compliance, or customization, then training your sights on meeting those criteria is a natural for private cloud services.
It might be that Private Cloud from a service provider is the right answer. “I’m a bit stubborn and mostly alone in this but I think it’s not cloud unless it is off-site,” says Monty Blight, VP of Product Management for Peak 10. “Cloud service providers offer Infrastructure as a Service either on a multi-tenant platform (Public) or on equipment dedicated to the customer (Private). What most call on-site “Private Cloud” is virtualization and automation on gear that is owned, housed and managed by the organization. Yes, it’s faster and hands off more capabilities to the user but I feel like that is the natural evolution of an organization’s technical capabilities. In the end, it doesn’t really matter other than to make sure communication is crisp. Our customers build and manage their virtualized platforms and use Peak 10 for Colocation and Support. Or, they use Peak 10’s cloud services. Or, and increasingly, they use both in a hybrid solution.”
So, you like your virtualization hypervisor. You’ve built up an impressive set of management tools. You’re proud of your customized solutions. Should you base your decisions on private cloud design and deployment on these factors? Probably not. On-premise cloud architectures should be designed for a standard set of offerings and high levels of automation, focusing on a smaller set of operating systems, scale-out applications and use cases. Only then can your private cloud deliver the flexibility, speed, performance and automation tailored for those use cases.
Consider a general-purpose solution that is ready to use when time-to-value is important; when customization and “getting it right” are more important and the skills are available, consider open-source frameworks and best-of-breed technologies.
But, again, deploy private cloud in an evolutionary way. Start small and expand standard offerings incrementally based on changing demands, with an eye to designing for interoperability down the road.
Let Transformation Begin
The transformation that a successful private cloud implementation can bring to life doesn’t begin with or in the cloud or with cloud technology. True transformation begins on a much bigger stage … by first transforming the organization’s people, processes and the organization itself. The right solution then follows.