While small and midmarket businesses gobble up public cloud services, enterprises are experimenting with private
cloud computing to gain efficiencies from existing resources. The exercise has resulted in some best practices -- as well as some pitfalls, experts say.
A private cloud takes advantage of in-house technology to provide metered services to users. Concerns such as dependability are on the rise, however, according to Frank Gens, senior vice president and chief analyst at IDC, a research firm in Framingham, Mass. While it's cost-efficient to share resources, some IT executives worry about balancing resource spikes in a shared environment, he said.
Regardless of such potential pitfalls, enterprises need to get on board with the cloud computing paradigm, said Steve MacLellan, senior vice president of enterprise architecture financial services at Fidelity Technology Group in Boston. "The horse is out of the barn and IT is playing catch-up," he said. "Think of how many people use WebEx and make extensive use of cloud apps. The No. 1 application in most enterprises is email, which is rapidly being outsourced to public clouds."
Enterprises inevitably will build private clouds to coexist with public clouds in a hybrid environment, experts said. Being aware of some of the snags CIOs might encounter when they build such environments will help them manage executives' and users' expectations. Here are three private cloud computing pitfalls to avoid:
1. Expecting speed and agility
Speed and agility are commonly touted as benefits of cloud computing, but they aren't when it comes to building your own cloud, according to Debe Gash, CIO of St. Luke's Health System, which operates 11 hospitals and several clinics in the Kansas City, Mo., metropolitan area. To comply with new regulations for electronic health records, she opted to use public cloud services from RelayHealth, based in Atlanta.
"It's been great," Gash said. "It's just so easy to implement; we're finding a huge win." And while Saint Luke's is well on its way to compliance, she sees her competitors stumbling with private cloud computing. "As I talk to my peers, we're so far ahead of them. They're building heavy infrastructure in their own private world and trying to connect, and they're not able to get it done," she said. "I don't think we could have accomplished what we have this year if we had to build it ourselves."
2. Waiting to get into the game
Gash makes a good point about the complexities of private cloud computing, but organizations that get caught in "analysis paralysis" will suffer even more than those that make the leap and build their own clouds, according to Brian Wilson, vice president of services and support at Surgient Inc., an Infrastructure as a Service provider in Austin, Texas, that has been involved with 150 deployments of private clouds for Fortune 500 enterprises.
"Cloud computing is what the Internet was 20 years ago," Wilson said. "It's a game-changer." Granted, Surgient benefits from the hype surrounding the cloud, but experts agree that the biggest downside is not doing any cloud development. Once private clouds are up and running, they are paying for themselves in one to two quarters, he said. By his estimates, one in 10 enterprise servers are sitting idle; private clouds enable them to move from less than 10% server utilization to 70%.
3. Failing to understand self-service and true production needs
You can look at self-service two ways: from the end user's point of view and from the cloud administrator's perspective. For the end user's point of view, consider an organization with finite resources and defined time frames, such as an airline or hotel, Surgient's Wilson suggests. Taking reservations is important, but guaranteeing them is critical for self-service. If the back-end process fails, the end user will go elsewhere.
From the cloud administrator's perspective, it's necessary to determine the right balance of automation and control, Wilson says. Some resources have a greater need to be deployed for certain groups. It's important that the administrator define appropriate policies to ensure that the resources are available to those groups, then let the private cloud run.
In the end, self-service becomes a moot point if the private cloud cannot reliably support mission-critical applications, isn't redundant, isn't properly monitored and does not live up to promised service-level agreements. Because clouds are almost too easy to grow, you should determine the scale points up front to avoid future headaches, Wilson said.
Let us know what you think about the story; email Laura Smith, Features Writer.