Why the cloud of today isn't the cloud of tomorrow

With stories about cloud technology success and business leadership emerging daily, it's tempting to say we understand everything about the cloud, how it will evolve and who the winners in the cloud space will be. But we need to resist that temptation, because we know very little about long-term cloud trends based on near-term cloud projects. Today's cloud, according to enterprises, isn't tomorrow's cloud.

For all the excitement about cloud computing, current spending on cloud is relatively small when compared to what will ultimately be spent. Judging the whole market based on what's happened so far is like picking a marathon winner 138 feet from the starting line. But knowing that it's premature to call a winner in the cloud space doesn't explain why that is, and understanding the "why" could be a help to cloud planners.

What is moving to the cloud now?
The first reason is more financial than technical in nature. Information technology investments are "capital investments" in a financial sense, meaning they're long-lived. On average, IT assets are written down over a period of about five years. If anything is rendered obsolete before that write-down period is ended, the balance has to be written off, creating additional cost and lowering companies' profits for the period. Not surprisingly, companies don't like to write off IT assets, which means they can't just massively shift to the cloud. Instead, they dip their toes in the cloud with early project activities that haven't yet been invested in. According to enterprise surveys, over 65% of all their current cloud computing commitments are either pilot projects, development and testing for new applications or simply cloud applications to support new business needs.

Development and pilot project work, according to the surveys, tends to consume Infrastructure as a Service (IaaS) offerings because it's the most flexible in hosting multiple applications running on a range of operating systems and middleware platforms. New business needs are most often fulfilled with tactical Software as a Service (SaaS) cloud procurements. Thus, current enterprise cloud spending is clustered at both ends of the spectrum of cloud services -- the most basic IaaS and the most integrated SaaS.

Where are the future cloud spending opportunities? According to enterprises, their largest cloud applications come from offloading mission-critical or core business intelligence applications from their own data centers during peak load periods or when an internal data center element fails. To make this offloading work, the cloud must host the offloaded application, quickly augment or replace the internal resources, and be able to access whatever data it needs to run. These key requirements aren't impossible to meet with IaaS or SaaS, but enterprises say they're difficult.

SaaS is an application-in-a-box strategy. Many enterprises' core business applications are supported by self-developed or customized software, so these enterprises can't adopt a SaaS cloud for application offload or backup. Where enterprises use off-the-shelf software, they find their choice is either not yet available in the cloud or available only from "risky" service providers lacking in security and availability guarantees.

For IaaS services to augment or back up critical local applications, the cloud host must have a complete image of the company's core applications, including a means of accessing data. Since it can take a while to commission a machine image on an IaaS, the machine image probably needs to be "hot." The IaaS service and the machine images of the application also have to be developed and maintained to support this fast-offload mission. Enterprises report that this process is expensive because it adds extra-cost features, like high availability and dedicated hosting to basic IaaS.

Judging the whole [cloud] market based on what's happened so far is like picking a marathon winner 138 feet from the starting line.

Where is the future of cloud?
Most enterprises believe their offloading missions for the cloud will require a form of Platform as a Service (PaaS). They also think that PaaS providers will have to offer Database as a Service to hold customer data. These DaaS services, hosted within the customers' data centers, will need to be accessible from the cloud in areas where it's impractical to migrate the full database or if it's prevented by security considerations or costs.

Enterprises also think that their "offload cloud" provider will offer presentation services, linking applications to the devices enterprises expect to use via virtual terminal support. These cloud presentation services, already available from some operators, will be used even for internally hosted applications. The PaaS offering will then switch work back and forth "underneath" the presentation cloud.

Many think that the most credible cloud partners are either not yet in the market or in the early stages of building out cloud services, further driving the future of cloud to a different place. Top candidates for enterprise cloudsourcing are their primary IT vendors (IBM, Microsoft and Oracle) and their primary network service providers (AT&T, Verizon, BT and other large national carriers). Enterprises believe the entry of these new players will change the market dynamic, both because these giants will define a new "mission-critical cloud" service set and because their entry into the market will change offerings of current cloud incumbents, including Amazon, Google and Rackspace.

Buyers define the market, and the cloud market appears to be headed for major changes.

Tom Nolle is president of CIMI Corporation, a strategic consulting firm specializing in telecommunications and data communications since 1982.

This was first published in July 2011