What’s with all the private cloud haters? They’re popping up like mushrooms after a spring rain – analysts, consultants, even an occasional peer on this site. In my opinion private cloud has a critical role to play. As with everything, it’s just a question of selecting the right tools for the job.
Private cloud skeptics, like ZDNet’s Phil Wainewright, rightfully question the role and impact of private cloud. However, type-casting private cloud’s “dreadful fragility and brittleness” shortchanges the transformational role private cloud can play in addition to its ability to drive real-time–to-market and efficiency benefits.
Instead of “either/or,” I believe it’s a question of knowing where and when to use the right mix of public AND private cloud technologies to deliver the services your enterprise needs.
Phil, like many pundits, cites the Netflix case study as a great example of an organization that’s successfully “burnt the boats” and moved wholesale to the public cloud. And it is!
The Netflix example is compelling but I don’t believe it reflects reality for most of today’s Fortune 500 IT professionals. These enterprises consistently tell me the same thing; that their future is neither exclusively cloudy or cloudless, but rather a hybrid of service delivery models – services that still need to be managed for availability, performance and against disasters across the entire end-to-end value chain.
So, instead of getting all wrapped up around what type of cloud, I believe we should focus on the real transformation – the transformation to delivering flexible, secure, shared services with transparent chargeback (or “show back” or “metering”) on consumption – not technology for technology’s sake.
Here’s a secret – your users don’t care…
One of the best places to get started with the shift in thinking to hybrid delivery is with services that are readily “consumerized” like smart phones and email. One of our Enterprise CIO forum members, Clorox, recently moved from mandating the device and service (in this case Blackberry smart phones and servers) to an open model, where the end-user selects their own device and simply connects to the Clorox email service (or “private cloud” if you will).
It’s similar here at HP too. Access to email services is all managed through a self-service portal and automated policy enforcement. I believe it offers a glimpse into how IT and end-users will interact in a hybrid delivery model. No matter what device, wherever I am, I get my messages. Is my HP email powered by a private or public cloud? I don’t know. All I know is that IT bills me monthly and it works. I love it.
Fixed vs. variable costs – FIGHT!
So if private clouds work so well for some, why all the heated debate? The technology’s essentially the same. The end-user’s experience is identical. So what’s the real difference? My take is that it’s a question of true variable vs. fixed costs.
At their core, the two flavors of cloud differ in that most private cloud implementations are provided with a high proportion of fixed costs. In the above example my mailbox is a variable cost for my business unit. (My business unit is only charged for each user.) But for IT and therefore our balance sheet, it’s ultimately a fixed cost. That is, IT and HP overall bears the real costs of licenses, hardware and data center space whether or not anyone uses the service.
According to industry estimates, 30-40 percent of a typical IT shop’s expenses are variable. If you’re a start-up or subject to large swings in IT utilization (say you’re running an online movie rental company that’s essentially idle outside of your peak load between 7PM and 11PM), then that’s way too low. The figure should be at least 60 percent. In fact, the higher percentage of your costs that are variable, the more control you have over peaks and troughs in demand.
For me, that’s the real difference between a private cloud and a public cloud. A private cloud represents predictable fixed costs. A public cloud represents variable costs. In the case of public cloud, when the service isn’t in use, the only person stuck with over-provisioned infrastructure is the service provider – and that’s a great topic for a different post.
So private cloud makes sense for frequently used services, or services where the peaks of consumption of one service coincide with the troughs of another to create a predictable “base load” of demand. Development/test workloads are one of the best candidates for this type of service and one of the reasons we developed our Private Cloud for Test software solution.
What about less-common services? The challenge from an IT perspective is when the users stop using the service. That’s where variable costs become important and where the public cloud makes the most sense.
Still, IT executives are wary. Frankly, I don’t blame them, for two reasons. First, there are a lot of vendors engaging in what I refer to as “cloud-washing,” in which suddenly everything they offer has been splashed with a cloud veneer. It’s tough to know when you’re looking at a truly transformative approach or legacy technology masquerading as a something new when it’s really not – you need to know what questions to ask your vendors to ensure you’re really getting what you need.
The second reason is a lot harder, the soft stuff.
Cloud computing forces you to make one of the biggest mental transitions since we moved from mainframe to client-server and client-server to the Web. It forces the shift from thinking about infrastructure and applications to thinking about processes and services.
These processes and services provide the foundation for a highly transparent cost structures, letting you run what you want where it makes the most sense economically. But true workload portability requires you to transform the way you build applications and the way your structure your operations to support them. It’s here that I think private cloud provides a great stepping-stone.
Private cloud, represent!
Closing out, while I think that private cloud is wrongly cast as a poor-man’s cloud by the haters, there is one area where I agree the difference is material - the economics: maximizing utilization and value-added is more involved than simply virtualizing servers. We as an industry still have a long way to go in reshaping infrastructure and software licensing to enable variable cost structures. (Before you ask why we’re not there yet, I know from experience, it’s far more complicated than you can imagine!)
So, while you can’t yet wave a magic wand and move to a completely variable cost model for private services, you can use private cloud to enable time to market, efficiency and compliance improvements while preparing your people and your applications for the brave new world of hybrid delivery.
What’s more important to you – getting your people and processes ready with private cloud or having the truly variable cost structures of public?