Cloud Computing Both More Agile and Less Expensive

January 26, 2012 Off By David
Object Storage
Grazed from Network World.  Author: Bernard Golden.

In Silicon Valley, the saying "it’s a dessert topping and a floor wax" is often used to puncture the pretensions of a product that promises that it can address every need; it’s applied to products claiming oxymoronic qualities. For example, the saying would be applied to a product that claimed to perform network management and word processing–two different, mismatched, and disharmonious functionalities.

I’m reminded of this when I listen to a debate common to cloud computing discussions. One person will assert that cloud computing is less expensive than the traditional IT infrastructure operations; a second person then says "cost isn’t really the issue for IT, what they really want from cloud computing is greater agility," meaning rapid provisioning of infrastructure resources…

IT Costs Matter

Those who claim that IT organizations aren’t really focused on cost, but are much more attracted to cloud computing’s agility characteristic, are guilty of wishful thinking. Oftentimes, this claim is stated by vendors that cannot compete on price with other cloud providers (that is to say, Amazon).

Cost and Agility Aren’t Mutually Exclusive

The attempt to downplay the importance of cost regarding cloud computing is misguided. However, there is a greater mistake in branding cost and agility as conflicting choices, with the implication that choosing one means forfeiting the other. The fact is that low cost and agility both depend upon the underlying foundation of cloud computing: automation. It is automation that supports both lower costs and agility, and both of them equally reflect its nature.

Cloud computing represents the automation of virtualization–the replacement of manual system administration by execution of scripts. By marrying script execution to virtualization, the provisioning process is streamlined and shortened in duration. Instead of the days to weeks a manual provisioning process can take, automated execution can accomplish the task in minutes. A side benefit of automation is that human error is reduced, so the process is typically higher quality than manual administration.

Once one steps back from the "less filling, tastes great" conflict between agility and cost, the transformational nature of this automation can be examined more clearly. In every other industry in which manual processes have been supplanted by automation, a number of things have occurred:

1. Cost for the good (or service) has dropped. One has only to look at the automobile manufacturing to see how mass production changed the end user pricing of cars. Henry Ford’s automation turned cars from a plaything for the rich to a essential of everyday life. The fact is that automation reduces the labor content of a process and pulls cost out of it. Consequently, given cloud computing’s automation, you can expect that the overall cost of IT infrastructure will decline, probably quite dramatically. Those who believe that "enterprise" pricing will continue despite the spread of cloud computing are misguided.

2. Incumbent providers have been confronted by the need to adopt the new technology or go out of business. The number of U.S. car manufacturers dropped by 90 percent in the quarter-century after Ford introduced mass production. Only those who could adopt Ford’s techniques and compete with him on price survived. This means that every IT organization and service provider should recognize what the future holds and develop a strategy to adapt to it.

3. Demand for the good or service has skyrocketed as it becomes cheaper. Henry Ford sold 15 million Model Ts, a number surpassed only by the Volkswagen Beetle and the Toyota Camry (both of which sold higher numbers in a market probably 50 times greater than the auto industry in Ford’s time). This means that cloud computing is going to cause the demand for IT infrastructure to increase enormously. We consistently recommend that IT organizations develop strategic plans that include a "10X" scenario, because–based on historic analogies–IT will soon see an enormously higher volume of demand.

4. Everyday life is transformed by this new, less expensive capability. The organization of American life today would be shocking to a citizen of a century ago. Every dimension of our lives assumes the easy access of mid- to long-distance personal mobility based on the car. We work in suburban office parks, date and marry people who live 50 miles away, think nothing of taking a weekend visit to a museum or amusement park 200 miles away. Once the easy availability and low cost of a technology is available, it is rapidly subsumed into environmental assumptions. With respect to cloud computing, clients are now planning highly scaled, short duration applications to meet temporary business needs–because they understand that inexpensive, easily available (i.e., agile) infrastructure is there for the taking.

Rather than the arid faux dilemma dialogue of cost versus agility, a more appropriate response is to consider what an IT world where both are considered prerequisites for commonplace infrastructure. Here are some thoughts:

  • Agility at the front-end isn’t enough. While agility is typically embraced by senior IT management trying to reduce developer dissatisfaction (and "shadow IT" use of public cloud computing), agility needs to be present throughout the entire application lifecycle. This means production environments, too. Failing to automate infrastructure throughout the lifecycle means that production environments will remain high cost (and error prone, to boot). The nature of applications is shifting to variable load, short-duration designs as application groups learn more about how to build and operation them (see #4 above for why this shift in design occurs). It’s crucial to support agility in every part of the infrastructure.
  • Infrastructure itself must be automated. Being able to provision resources for an application is insufficient; the growth in demand, large-variation loads, and increase in application scale means that overall infrastructure demand will grow. IT Infrastructure and Operations must be able to handle the shared resource pool as deftly as applications handle the resources used at the application level. If this sounds like "DevOps," you’re spot-on. Whether you like the term or not, the implications of cloud computing demand infrastructure agility.
  • Price and cost benchmarking will be part of the daily environment for IT organizations. Offering rapid resource provisioning with the belief that it obviates any requirement for cost competitiveness with external providers is delusional. Putting agility into place and failing to meet market pricing will elicit a "what have you done for me lately" attitude from business units and application groups. Switching costs for users are vastly lower in this world, and transparency of pricing implies that benchmarking will be a fundamental fact of life for IT.

Don’t get trapped in the "either-or" discussion of agility versus cost. Don’t even imagine that these two characteristics are really different. They are just two expressions of the same phenomenon, virtualization automation. One might say they are two sides of the same coin, except that intimates a "heads or tails" reality. Cloud computing is more like a "heads *and* tails" story.