My morning coffee reading today was Josh Stephens’s “Top five virtualization problems: Building a private cloud” over at TechTarget SearchNetworking. I was all set to get some good advice on private clouds. Unfortunately, I stumbled on another Cloud Myth.
The article starts off well enough, but runs aground when it gets to what should be the meat. Josh writes:
Ultimately, the private cloud is nothing more than an evolution of your own data center and the way that you think about compute resources. There really isn’t a finite line you cross in order to become a private cloud — except maybe the ability to provide the following features:
- Elasticity of resources
- Ease of provisioning
- Accounting of use
In cloud nirvana, departments within your organization would be able to automatically provision their own computing resources, getting more or less as their requirements demanded. Their usage would be tracked and automatically billed back to their department. Your infrastructure would self-optimize based upon available capacity and requested computing needs. That, at least, is what most people think of when they think of private cloud. However, most of us will never see that sort of environment and many of us never want to.
My alarm bells went off big-time when I read the first sentence quoted, above: “Ultimately, the private cloud is nothing more than an evolution of your own data center and the way that you think about compute resources.” Josh never really explains what “evolution” he might have in mind. I suppose that means the addition of some tools, products, or technology to your environment, but I’m speculating on that. The thing that got me off the couch to write this blog is the second half of that sentence: “…and the way you think about compute resources.” Clouds are far from just a state of mind. It isn’t as if I wake up one morning, read a technology blog, decide that my organization needs a cloud, close my eyes and change how I “think about compute resources,” and I’m done. There’s a bit more to it than that.
The last sentence in the quote is also a little disturbing: “However, most of us will never see that sort of environment and many of us never want to.” I guess my questions would be, if you don’t want to see that sort of environment, why would you be building a “private cloud” in the first place? And what would that private cloud look like if it didn’t include attributes similar to that? A private cloud is not simply “virtualization on steroids” or worse yet simply “virtualization with a different state of mind.”
At ServiceMesh, we try to reiterate to our customers that building and running private clouds is really about the intersection of multiple organizational changes:
- Operating model
- Supporting technology
Let’s take them in order.
Strategy — ServiceMesh firmly believes that cloud computing as a whole is a game-changing technology shift, akin to the rise of the Internet. Firms that recognize this will be able to gain significant competitive advantage over firms that don’t. But either way, clouds are NOT just a new technology that can be blindly rolled out by the IT department alone as “virtualization, phase 2.” As a firm, you’re going to want to make sure that your cloud strategy supports your firm’s business strategy. CIOs, I’m looking at you here. This is your opportunity to show that you are a peer with all the other “chiefs” in your firm, not just the guy responsible for keeping the email system running.
Operating Model — Once you have determined your strategy, you can work through the operating model that will support that. A cloud operating model is different than a typical IT operating model because it focuses on rapid, direct, self-service access to IT resources by business units. The IT department has a strong role in setting up the processes and systems that make this possible, but generally speaking, IT’s main goal is to get out of the way. The various technical capabilities that Josh writes about (self-service access, usage tracking, and billing) are important tools for implementing the cloud operating model, but there is more to it. For instance, what about approvals? If you lard-up your operating model with lots of approvals, you’ll end up slowing things down to a crawl, negating all the agility benefits you might have been hoping for. What about governance? If you allow everybody to access anything, at any time, anywhere, you might end up paying fines for breaking laws in one or more countries. That’s particularly true for a large multi-national firm that does business around the world. Understanding the new operating model that you’re trying to implement is imperative, and ideally your operating model should be designed to support your firm’s business strategy.
Supporting Technology — Deliberately last and least, we have the supporting technology. Private clouds definitely need supporting technology to implement the operating model that you come up with. All the things that Josh listed are good candidates for inclusion, but you’ll want to think them through and make sure your choices are supporting the operating model and business strategy you have designed. Call us at ServiceMesh and we can help you work through all the pieces that you’ll need to have. You’ll definitely want the ServiceMesh Agility™ Platform to help you govern your private and public cloud environment, but there are a multitude of other pieces from a variety of vendors. We have the experience to know what is both cost effective and what works, increasing the odds of success of your private cloud project.
Interestingly, the first two items in our list, strategy and operating model, are largely independent of any specific cloud. If you’re a firm of any reasonable size, you’ll want to design a strategy and operating model that works for a hybrid cloud environment, both private and public, internal and external.
In summary, private clouds are not just a state of mind. You can’t just think about your computing resources differently. Rather, you’ll want to think through the strategy and operating model you’ll use within your firm, and once you have done that, acquire appropriate supporting technology to implement those.
You must be logged in to post a comment.