I always shake my head when I hear someone say that private cloud computing won't take off because enterprises can't realize the economies of scale that a public cloud provider can. Thing is, the benefits that enterprises get from a private cloud are not the same as a public cloud. I was reminded of this while reading F5's Lori MacVittie succinctly sum up the difference in goals between private cloud and public cloud adopters in Focus of Cloud Implementation Depends on the Implementer: "Private cloud implementers are not trying to be Amazon or Google or Salesforce.com. They're trying to be a more efficient, leaner version of themselves--IT as a Service." While all organizations want to be efficient, the way they measure their success is vastly different from how cloud providers measure success.
One of the benefits of private cloud computing is operational efficiency--which is perhaps the benefit--the ability to do more with less, more quickly and with fewer errors. It's that simple. Unless you are a really, really big IT shop, you probably aren't going to achieve the economies of scale that big cloud providers achieve, but, as Lori points out, that really isn't your goal, is it? Your company isn't likely trying to be the next Salesforce or Amazon Web Services. But if you aren't seriously examining how to leverage automation and more flexible data center architectures, then you are doing yourself, and your company, a disservice and putting off the inevitable. You can automate IT and you can't avoid it.
This comes on the heels of sitting down with HP in Santa Clara and getting a demo of the management software for their A series switches. We got to talking about software deployment, application templates and the notion of IT self-service. Providing a menu of IT services such as Wiki, Database and SharePoint to users is neat, but I don't think that is a terribly realistic goal. Applications need a bunch of customization once the core systems are installed. Unless you are in a really big organization that spends a lot of time spinning up applications for users and you can thus justify building a self-service portal that will provision applications that are ready to serve users, the idea of giving business users IT self-service is probably not practical.
However, you can be more efficient and deliver services faster if you automated 90 percent of the application deployment process and then simply had to put the final touches on the application before shipping it to the business unit. If I focused on one install, it would take about 3 hours to install an instance of SQL Server, and that includes installing, configuring and patching the OS, as well as installing, configuring and patching SQL Server. It doesn't include downloading the bits--I had those stored locally on a SAN. I am not done yet, however, I still have to install, configure and patch the application that will use SQL Server, and then put them together, add them to DNS, etc., etc. I can speed this up by cloning a virtual machine, but what I really want is to fill out a form with basic information, press "go" and then do something else while the self-service system does the rest. That's an IT self-service concept that I can get behind.
In fact, as far as naming your internal data center as a private cloud, I'd suggest, as the Enterprise Strategy Group's Steve Duplessie did in January 2010, that "there is no such thing as a private cloud. A private cloud is called IT. We don't need more terms for the same stuff."
