The cloud computing model is rapidly working its way up the application stack.

The latest example of innovation surrounding a cloud computing model is a company called Elastra, which just picked up another $12 million in funding thanks to Amazon and Bay Partners, a venture capital firm based in Menlo Park, Calif.

What makes Elastra different than your average hosting company is that it provides a virtual layer of database and application server that customers can dynamically invoke as applications loads increase or decrease. That's different than the typical capacity-on-demand models usually associated with hosted computing offerings.

Elastra at present is basing its service around open source technologies such as MySQL, Postgres and EnterpriseDB, but it's only a matter of time before they apply the concept to more proprietary middleware frameworks.

What makes the model attractive in this case is that it reduces a lot of the risk associated with launching a new application. You don't have to first invest in a lot of infrastructure on the assumption that that application will be popular only to discover that people have no interest. Conversely, if the application proves popular, scaling it out is only a matter on adding additional compute units.

Obviously, Amazon likes the concept as an extension of its EC3 storage-on-demand services. But the real challenge here is whether software vendors are prepared to sell and market their software using this type of model. SAP, Oracle and IBM, largely due to their own self-interests, have all been pretty slow when it comes to embracing cloud computing as a model for delivering software.

This may create some significant opportunities for start-up companies to deliver rival offerings using a cloud computing model that has a lot of upfront economic benefits for the customer over the models currently favored by established software vendors.

The whole enterprise licensing model today is pretty much a mess no matter how you look at it. But the rise of services such as Elastra only make it inevitable that they way we consume and pay for enterprise software is going to change.

This doesn't mean that every piece of enterprise software is going to move into the cloud. But it does mean that the cloud is gong to be an extremely important conduit for how we consume software. Ultimately, this will lead to a blended computing model that meld software components residing locally on premise and in the cloud into a set of dynamic services that should ultimately prove to be 100 times more flexible than any thing we seen to date.