We won’t have true cloud computing until we have a services-based infrastructure and standardization of cloud management frameworks.

We may call it “cloud” today, but what we really have with the offerings today is “capacity on demand.” We don’t actually have all the pieces necessary to execute on the vision that is “cloud computing.”

We’ve almost completed server standardization through virtualization but we haven’t really begun to standardize network and infrastructure services. And we’re certainly nowhere near ready to standardize on the cloud and application frameworks that will enable a seamless Intercloud. The term “utility” has many meanings. One of them is an “economic term referring to the total satisfaction received from consuming a good or service.” The utility of cloud computingcompute on-demand services today is fairly middling on the “W00T” scale from an enterprise consumer perspective.

On a scale of 1 to 10 I’d say we’re at about 3 today. We’re at compute resources as a service, at capacity on demand, but we’re not at infrastructure as a service. We’re not even really close yet.

Public cloud computing today is about compute services. Very few public cloud providers offer any kind of Infrastructure services over and above (or under) load balancing, compute, and storage. The nomenclature used to describe these offerings, Infrastructure as a Service, is far grander than reality.

Werner Vogels on behalf of Amazon recently dismissed private cloud computing:

Speaking at CeBIT 2010, Amazon vice-president and CTO, Werner Vogels, praised public cloud computing in terms of its scalability, cost savings, security and reliability.

He compared private clouds, which are owned and operated internally by businesses, to generators used by private companies to make electricity before the availability of public utilities at the turn of the century.

His comparison is completely unrealistic, as are the myriad other comparisons of public cloud computing to electricity. If perhaps public utilities generated different types of electricity based on specific needs within the consumer’s home then the comparison would be a fair one. But public cloud computing, like utilities, is generating and delivering plain old vanilla compute resources. It’s not infrastructure as a service, it’s compute resources as a service.

Vogels comparison focuses primarily on costs and while accurate when comparing the costs to generate general-purpose compute resources, fails to take into consideration that well-established enterprises have built not only a robust application infrastructure but a cooperative network and application delivery network infrastructure through which applications are secured, delivered, and scaled. Public cloud computing offers only scalability at a lower cost.

Paul Fremantle, CTO of WSO2, offers a reminder of the stark reality behind enterprise application architectures in “Platform-as-a-Service freedom or lock-in”:

But Enterprise Architectures are not just about Java code: what about an ESB? What about a Business Process engine (BPMS)? What about a standard XACML-based entitlement engine? So far PaaS has generally only addressed the most basic requirements of Enterprise core services: databases and a identity model.

This is the same question (and underlying lament) of those considering the use of public cloud computing: Where’s the infrastructure as a service? Where’s the firewalls? Where’s my network systems management integration? Where’s my Load balancer? My application acceleration? My cache? My WAN optimization?

Where’s the freaking beef?

The beef is still in the corporate data center, in an environment in which organizations can leverage and control infrastructure and application infrastructure resources that simply aren’t available today in public “utility” cloud computing offerings. It isn’t that organizations aren’t taking advantage of the cost savings Vogels repeatedly calls out – they are. But they are doing so on a case-by-case, application-by-application basis. Applications that aren’t tied to existing infrastructure, that don’t rely on myriad other infrastructure and application infrastructure services can and are being deployed in IaaS systems like Terremark, like GoGrid, like BlueLock, like RackSpace, like Amazon EC2.

Applications that require only general-purpose compute resources are ideally suited to public “utility” cloud computing environments and, in fact, IT ought to refuse to deploy such applications locally based solely on the cost savings Vogels mentioned more than once in his CeBIT presentation. But for highly integrated, mission critical applications that rely on a variety of application and network infrastructure services for security, for performance, for availability, for fault-tolerance…those are not general-purpose compute resources. They are in fact specialized compute resources that are in general not available in the public cloud today.

Vogels’ statements are not in an of themselves inaccurate, but he neglects to recognize that some functions in a data center are not considered “utility” and are actually inherently important to the successful delivery of the organization’s applications and are required. Until public cloud computing can generate some specialized compute resources in the form of required application and network infrastructure services, such environments will remain, unfortunately, little more than capacity/compute on demand.

Whether or not we need cloud – true cloud – is a completely different discussion. For many organizations capacity on-demand is all they really need and want anyway – making today’s “cloud” perfectly suited for their purposes.