As usual, the hype has gotten way ahead of the reality. This time, I’m talking about cloud computing.

Those “pure play” companies with a vested interest in the success of the cloud will point to what they call the growing number of large enterprises that either are experimenting with cloud computing, or have already moved a pilot project here or there onto the cloud. And certainly, more software, platform and infrastructure suppliers are moving their offerings into a cloud-ready, subscription-based solution that they believe will be poised to reap the benefits of this shifting development and deployment model.

Yet there are still the same problems that stifled the adoption of the Application Service Provider model a decade ago: security, dependencies and control. On top of that, add multi-tenancy, scalability, management, governance and provisioning of customers, and you can see why the earliest cloud movers are investing in private clouds. (This, to me, seems a misnomer anyway. It should be called “efficient computing,” as most definitions I’ve heard of private clouds talk about optimizing utilization of servers to reduce both their number and the costs of IT—which you still own and pay for, by the way.)

In fact, Sinclair Schuller, CEO of SaaSGrid maker Apprenda, sees the cloud bubble starting to deflate in 2011. “The level of education around the cloud has increased so much in the past 12 months that cloud vendors will be called out for what they do and do not provide,” he said. “The hype cycle is broken, and we’ll see a backing off of throwing the term ‘cloud’ around.”

Al Aghili of Managed Methods, which this month will release cloud-based middleware governance as a service, said the cloud has not yet reached a tipping point. “People are asking about it but are far from deploying in production,” he said.

Right now, there are three aspects to the cloud: software-as-a-service, platform-as-a-service, and infrastructure-as-a-service. The first is way ahead of the others at this point, although worker productivity software beyond CRM systems is only now making its way online.

“The first cloud applications are 10-year-old designs. They’re not composite, they’re not using business process flows or ESBs,” said Paul Fremantle, cofounder and CTO of SOA middleware provider WSO2. “The first steps in the cloud are things that have been easy to contain by using a PHP or Ruby runtime, or Tomcat. That will have to change significantly in the next couple of years.”

Several platforms have already been developed, such as Amazon EC2, Google and Salesforce.com, but those are proprietary. Apprenda’s Schuller sees a revolt against vendor lock-in in 2011, a “backlash to Benioff’s handcuffs,” he called it, referring to Salesforce CEO Marc Benioff.

Fremantle said that the people trying to win the platform story are “trying to tie people into the Web. The platform is owned by them, ran by them, and there’s no moveability. It’s Google or nowhere. Now that’s fine if you like Google, but it’s kind of a dangerous story,” he said.

Perhaps the greatest amount of work needs to be done in the infrastructure area. Compliance, for example, is tougher in the cloud, when the services you need to monitor are not your own. “Traditional compliance tools that look at logs don’t work well in the cloud,” said Aghili.

Integration with back-end systems is another issue where the cloud acts as another layer of indirection. Services will need the ability to punch through firewalls of entities with which an application has integrated, and monitoring and governing that creates problems as well.

“These kinds of integrations have to be provisioned and monitored correctly,” said Larry Alston, president of FuseSource, which sells open-source middleware. “If you can’t drop a service, or a virtual machine, into an environment and manage it, there will be problems.”

It’s critical that the cloud evolve in the right way. For that to happen, cloud providers are going to have to invest in the plumbing and tools that are based on loosely coupled architectures designed for utility computing. Trying to move existing platforms to cloud by opening up some APIs doesn’t take into consideration how people want to use software in the cloud or how they want to provision servers.

Let the hype die down, and let the smart folks get to work so the reality can catch up to the promise.

David Rubinstein is editor-in-chief of SD Times.