Monday, December 20, 2010

Private Clouds: First, There Is a Mountain, Then There Is No Mountain, Then There Is…

One of the great things about having savvy, articulate friends is that I can occasionally appear savvy and articulate (at least a little) merely by commenting on what those friends have to say. Who could resist? Not me, certainly…

My learned industry colleague and friend Peter Coffee of recently opined that the idea of a private cloud – a cloud computing infrastructure owned and operated by and for a particular company – is a choice that doesn't really exist. If you own and operate the infrastructure, it ain't cloud computing in its most true sense, Peter said. (Of course I'm paraphrasing. You can read his exact words at

But another learned industry colleague and friend, Andi Mann of CA Technologies, has opined recently that the public cloud – THE cloud, according to Peter and many others, I'd wager – is not for everybody, and certainly not for every business or government agency. Andi makes many cogent and salient points, which could be taken in summary as an argument in favor of private clouds. (You can read Andi's exact words at

The thing is, I agree with Peter and with Andi. At least partly.

I think the core issue here is a need to,  as they say in parliamentary procedures, "move to divide." There's the issue of IT service delivery, which is separate from IT service consumption.

If I own an IT infrastructure and I configure and manage that infrastructure in a converged, unified way, I can deliver services that can be consumed "by the drink" or per user/per month. So to my users, it looks a lot like cloud computing. Users get authorized and simply use the services they need, as they need them. But what I've built and am operating isn't THE cloud, and may or may not be a cloud. It's what some savvy vendors such as Egenera and more and more savvy analysts describe with terms such as "unified computing" and "converged infrastructure."

Make no mistake – converged infrastructures are incredibly valuable, especially if and when they help companies to manage IT more efficiently and economically. But there's nothing written in stone that says a converged infrastructure has to result in cloud-like on-demand service delivery or consumption. Convergence and unity can, at least metaphorically, stop inside the data center door and still help to reduce operational costs, improve operational responsiveness or both.

So, as Andi Mann argues, not every business can or should make the wholesale leap to public cloud solutions. However, as Peter Coffee said, it's not clear that such businesses need, want or even can build private clouds. So what do business decision makers really need and want?

What many need is a set of effective processes for evaluating and comparing current and candidate solutions. Those processes should be used to decide if, when and where it makes sense to adopt and integrate cloud-based services into incumbent environments. (Maybe we can call such adoptions and integrations "cloudbursts." Maybe not.) Those processes can and should also be used to decide if, when and where it makes sense to deliver on-demand utility-like IT services to users. Whether those services originate from premise-based, cloud-based, physical and/or virtual computing, storage or network platforms.

With such processes in place, business and technology decision makers can collaborate to evaluate, compare and select the best available service and resource management solutions. These, in turn, will help businesses to deliver consistently efficient and economical services to users, again wherever those services may reside. (This is why Network World recently opined that private clouds are "not for the faint of heart" in its comparison of five cloud management solutions, as you can read at

If you're at a company that's serious about building a private cloud, you should look at the Network World comparison. You should also look at what analysts and users are saying about how Egenera, Cisco, HP, IBM and other vendors are approaching the growing need for converged, integrated management of physical, virtual, premise-based and cloud-based resources. I think this is the real goal of many if not most efforts focused on private clouds. I also think that "private cloud" is an unfortunate term that is likely more helpful to vendors trying to sell stuff than it is to business decision makers trying to run their businesses better. But I don't think the term or the debate over its definition and validity is going away any time soon…


  1. and last would have attained enough achievement with this process. thanks dortch . I recently launched website cloud computing security awareness

  2. Great post. It's interesting how while cloud computing seems to be the buzzword of choice these days, ambiguity around what is and isn't cloud can lead to some very heated arguments. Particularly for those who have been around long enough to have seen the same concepts recycled a few times before(HPC, ASP, unified computing, etc...)

  3. Finally! Somebody counters the “there is no such thing as a private cloud” discussion with a cogent and insightful argument.


Comments welcome!