An IT buzzword since IBM first heralded it in 2002, on demand computing is drawing serious attention—but what exactly is it and what value can it deliver to businesses?
On demand computing is the provisioning of computer-based resources to enterprises as they need them, whether these resources are servers, networks, communications bandwidth, I/O, or software.
Some enterprises see on demand as utility computing, where IT rents servers and storage as commodities from third-party providers—and stops renting when demand wanes. Others see on demand as grid computing, which is frequently used in scientific and academic communities to tie together thousands of computers along a network, building them into a super computer for problem-solving. Still other enterprises see on demand as the renting of vertical industry software applications from outside providers. Finally, to many, on demand is simply a new name for the age-old challenge of effectively allocating and deallocating internal technology resources to meet the fluctuating demands of business.
“There’s a lot of confusion in the market about on demand,” says Ben Pring, vice president of Research for Gartner. “Some define on demand as Software as a Service (SaaS). Others call it an Application Service Provider (ASP), a host, utility, or grid computing. The vendors themselves have contributed to the confusion because their initial definitions of on demand were very broad. This was understandable, but it has led to downstream complications in the marketplace that make it difficult to determine what’s really happening.”
External On Demand Deployment
Definitions of on demand may differ, but there are really only two ways to deploy it: internally and externally. Enterprises have been reluctant to pursue either aggressively.
“There’s one exception in the external on demand area,” said Gartner’s Pring. “The market that has really taken off is SaaS. This is where an application is owned and maintained by a third-party provider with expertise in a particular niche. An example is salesforce.com. Customers pay on a subscription basis and salesforce.com takes the responsibility for managing the hardware and the application. We estimate the SaaS industry at $5 billion worldwide and it’s growing 25 percent per year.”
Pring says there are sectors such as public service and education that use external on demand computing—but maintains that the external on demand market hasn’t exploded as vendors thought it would. There are several reasons:
- Enterprises are uncomfortable with housing mission-critical data and applications with third parties.
- An enterprise has to install and configure systems to send them to outsourcers. This is time-consuming and difficult if you’re in a spur of the moment on demand situation. There are some applications specifically designed for scaling to third-party providers in times of on demand resource extension, but these applications are few.
- Outsourcing in on demand scenarios requires corporate IT to closely monitor and manage the appropriation of external computing resources; to actively manage the outsource vendor; and to continuously oversee that any computing resources purchased are delivering value to the business. On demand computing that uses external service providers is consequently best suited to enterprises that have a keen understanding of what they’ll get out of it, accompanied by a strong internal management methodology for on demand outsourcing.
- Often, IT doesn’t yet have key enabling technologies in place to facilitate on demand outsourcing. One of these technologies is virtualization, which literally allows a site to send a system image over the wire to a third party without having to hard-wire a remote operation.
“Using external third-party service providers for on demand capacity needs will be a gradual process,” says Scott Crenshaw, general manager of Enterprise Linux for Red Hat. “The primary impediment to this is the lack of universally available technology that will allow this to happen. . . . The first gate is the deployment of virtualization technology within companies. Once this occurs, it’s relatively no cost to the enterprise to take advantage of third-party resources in an on demand environment. The second gate is information that organizations don’t want third parties to have.”
Internal on Demand Deployment