I expect cost is one of the biggest reasons a lot of companies have not started building private clouds. The problem, I think, is that most companies assume it's just not a cost-effective use of their resources.
It's true that a private cloud may not always be the cheapest option, but it's definitely the most cost-effective way to build a truly reliable and scalable IT infrastructure. In fact, scalability is the key to a private cloud's value: The more you increase server capacity, the more cost-effective your router, firewall, and other networking systems get, and the more you reduce your server redundancy costs.
I won't go into an exhaustive study of the various costs -- and cost savings -- associated with private cloud computing. This is an area with a lot of established research, and a quick trip to a source like VMware Inc. (NYSE: VMW) will serve up some good case studies dealing with cost and ROI. Instead, I want to focus on a few issues that often get overlooked or downplayed when companies think about the costs and benefits of building a private cloud.
Hardware and software costs. A typical company already has most of the hardware components it needs to start building and running a private cloud. That's because there's really no such thing as "cloud hardware," although there are options (such as enterprise-class, chassis-based systems) that are more cost-effective and probably necessary for a production cloud for scalability reasons.
For the most part, however, a private cloud is defined by the software and configuration choices your company makes, not by the hardware. At Fetch Technologies, our own private cloud solution is based on VMWare's vSphere Essentials kit and the Platform Computing ISF private cloud management software -- both of which probably cost much less to license than you think. We also use high-density appliances such as Juniper Networks Inc. (NYSE: JNPR) SSG integrated systems that combine router, firewall, and VPN functionality into a single appliance, allowing us to make very cost-effective use of limited data center space.
Infrastructure costs. This is worth discussing because it represents such a huge waste of resources. I once worked at a company that spent more than $500,000 just to provision the cooling and power systems for a server room -- and that didn't even include the cost of the servers. And it was still exposed to power outages because the building it was housed in wasn't capable of supporting a diesel backup generator.
That server room was practically useless -- and a private cloud architecture makes these kinds of planning and infrastructure disasters easy to avoid.
Staffing and IT productivity. Here at Fetch Technologies, we have one cloud administrator supporting 250 virtual servers. Compare this to my previous company, where six people managed 500 physical servers. And with end-user provisioning and management, we're still cutting days or weeks off service requests.
Scaling and redundancy. How big do you want a data center to be? I think there's a simple answer to this question: How many servers do you feel comfortable losing at any one time? Smaller, faster, cheaper data center designs -- the hallmark of a private cloud architecture -- make it possible to build more scalable, redundant IT infrastructures. Most companies that take the opposite approach, concentrating growth in a single data center, never achieve true redundancy -- it simply becomes too expensive to duplicate.
Your own organization's mileage will vary, of course, in terms of what a cloud computing solution costs you to build and operate. Cloud computing isn't necessarily cheap, but it is more cost-effective in the long run than a traditional IT infrastructure. And if you're focused on the big picture, that's what really matters.