It seems like every day now we hear about a new cloud-based service or Company X moving to the cloud. Microsoft and Google are locked in a legal battle of epic proportions over who will get to provide cloud services to the US Department of the Interior for a contract estimated to be worth over 50 million dollars, more if they can lock-in renewals. And with this shift in how we build software for the internet has come the stories of failure. Within a two week period in April 2011 there were no less than four major failures of cloud-based services from three separate providers. All those failures were so severe that the companies involved were required to provide recompense for breaching their unrealistic Service Level Agreements.

CEO's everywhere, entranced by high-priced salesmen armed with slick sales presentations; are moving everything to the cloud as fast as they can. They’ve been promised that the cloud will provide more capacity and better stability than their own data centers can possibly provide due to “economies of scale” or “core competencies” and a host of other industry buzz-words that these cloud providers claim to be able to provide. For most companies, whose primary business has nothing to do with technology, IT services are a cost center, a number on the balance sheet to be reduced at all costs, a necessary evil required to compete this technological era. The cloud is an easy sell.

Unfortunately, there is trouble on the horizon. Aside from the technical failures that are already being experienced, there are a number of other problems with the cloud. The first is data security. Security Auditors are now starting to recommend that all data that is legally required to be secure be stored internally, because, the cloud providers cannot prove that the data is actually deleted when the company asks for the deletion.

The second problem is purely physics. It is a problem both institutional stock traders and video-game players are intimately familiar with. That problem is latency. Latency is the silent threat to the cloud. As consumers demand ever increasing interactivity from the software and services they use, the cloud will become increasingly unable to deliver the performance required. User interfaces are most susceptible to latency because the human eye can perceive delays much smaller than the internet can physically deliver and the cloud, as currently constructed, requires that the server control everything about the state of software.

Many software developers have grown up in a world where there have always been computers. They have never been trained in anything related to the hardware that their software runs on. They do not understand the incontrovertible limits that are imposed on them by physics. They assume that the internet will always get faster, simply because it always has. However, this is not the case. The speed of light will remain the fastest that information can be transported and latency will always be a problem, because light, while fast, is not instantaneous.

The final problem is the technologies themselves that are used to build the applications and services that run on the cloud. HTML was originally invented in 1980 to provide a standard document format for universities to share research papers over the fledging internet. It was never designed or intended to create interactive user interfaces. Even with the use of JavaScript and technologies like jQuery that allow for the modification of HTML on the fly, HTML is not capable building the rich interfaces that consumers are already starting to demand due to its limited layout capabilities and lack of uniform display standards. Many web developers spend most of their time trying to convince a browser to display the site the same way as another browser.

In addition to the inability of HTML to adequately build interactive user interfaces the current web technologies are also unable to make use of the full capabilities of the consumers’ device. Some attempts have been made; many new browsers make use of the hardware for rendering and HTML5 includes support for some of the elements of a user interface and there are projects underway to bring 3D graphics to the web, but these solutions are doomed to failure. In the end, a document markup language cannot support the rich content made possible by these new capabilities on the consumers' device.

HTML was never intended to be used in the fashion that is today. The additions made to it recently are attempts to make new resources and capabilities fit into an old model. Its syntax is both limited and ill-suited to the tasks demanded of it. The security model is fundamentally flawed. And the limitations of the internet will prevent it from ever being truly interactive.

Learn more about The Vision ▶