Does the 'private cloud' make sense?

Go ahead, roll your eyes -- you've probably had enough cloud nonsense to last a lifetime. But automation is a good thing

Last week Jason Bloomberg of ZapThink wrote a funny, cutting piece that was meant to rip the notion of the private cloud to shreds.

I actually found myself agreeing with some of it -- but not with the blanket conclusion that "private clouds suck."

[ In the data center today, the action is in the private cloud. InfoWorld's experts take you through what you need to know to do it right in our "Private Cloud Deep Dive" PDF special report. | Also check out our "Cloud Security Deep Dive," our "Cloud Storage Deep Dive," and our "Cloud Services Deep Dive." ]

Let's take Bloomberg's points one at a time. His first is that people associate "the cloud" with public cloud services, where customers pay as they go for the resources they use rather than investing in infrastructure. With a private cloud, the customer necessarily owns the infrastructure, plus private cloud management software on top of that, so it's a capital expense like anything else you own rather than rent.

OK -- but isn't that why it's called "private"? If you don't get that distinction in a minute or two, maybe IT isn't the career for you.

A better argument can be found in Bloomberg's main point: In a private cloud, you'll never have anywhere near the ability to accommodate spikes in demand as with a public cloud, because public cloud facilities are massive pooled resources and private clouds are only as elastic as you're willing to invest in capacity to begin with.

True enough. Although I often hear about repurposing dev and test capacity to meet spikes in demand for production Web applications, Bloomberg is right. There's always a hard limit to scaling the private cloud.

His related point on this score is also correct: Because of that limit and the fact that the owner of the private cloud is typically the only tenant, the economies of scale running a private cloud can never match those of a public cloud.

Finally, Bloomberg makes a subtler point: Clouds of any stripe tend to require homogenous infrastructure. Can you say "rip and replace"?

When a cloud is not a cloud
A number of IT people I know who are exasperated by business management pestering them about "cloud this" and "cloud that" would agree with Bloomberg's arguments. But as with so much confusion in IT, I think the problem Bloomberg really identifies is one of nomenclature.

The public cloud and private cloud share a key characteristic: server virtualization at scale. No other technology has provided the data center with greater cost savings (in terms of increased hardware utilization) and agility (in terms of moving and scaling workloads) than server virtualization. But at a certain point, perhaps when you reach hundreds of physical hosts and thousands of VMs, virtualization becomes hard to manage with default virtualization management software.

By then, you need private cloud software to allocate resources and costs back to line-of-business stakeholders. And you want to take advantage of the agility of a virtualized infrastructure by allowing those stakeholders to provision their own resources -- self-service, in cloudspeak. Plus, as you converge data and storage networks, you'll want to virtualize storage and network resources along with servers. The software to do all that has arrived from or is being developed by VMware, OpenStack, Microsoft, Eucalyptus, and others.

While it's true that homogenous infrastructure is part of the deal, many, many enterprises have been building out largely homogenous virtualization farms for a while. It doesn't happen overnight, but as hardware is replaced, it tends to become part of the pool of virtualization hosts, where workloads can be scaled and moved around dynamically.

As for cost, although Amazon, Rackspace, HP, and so on may make good homes for temporary workloads, you still come out ahead if you buy and operate your own hardware for workloads you run all the time. It may be cheaper to run a huge cloud than a small one, but cloud providers are in the business of making money, and they're not going to pass all those savings to you.

As I've said before, in the future, commodity public cloud services may well offer a better deal to customers than maintaining their own data centers. That's the case even today for certain small businesses. But in the near term, the private cloud makes sense. Put it this way: As long as you intend to run your own data center, wouldn't you prefer to automate it as much as possible?

Bloomberg may be right: That rather modest goal may not be worthy of the "cloud" moniker. But given the choice, most IT folks I know would say yes to a more agile, efficient infrastructure. Plus, as I noted a couple of weeks ago, public cloud providers are aligning with compatible private cloud solutions, which should make it ever easier to move workloads between public and private clouds.

This article, "Does the 'private cloud' make sense?," originally appeared at InfoWorld.com. Read more of Eric Knorr's Modernizing IT blog, and for the latest business technology news, follow InfoWorld on Twitter.

Recommended
Join the discussion
Be the first to comment on this article. Our Commenting Policies