As IBM continues to push its on-demand computing initiative, it faces a number of hurdles, but the biggest may be the need to convince businesses to buy into it.
Computing on-demand, in its simplest form, is a strategy intended to deliver computing resources the way a power utility doles out electricity. When a network has a surge in demand, intuitive architecture triggers other resources into action, including idle servers, applications or pools of network storage. Companies pay only for the amount of time they use the services. On-demand also reaches into the realm of virtualization and policy-based management.
This year alone, IBM has brought to market at least a dozen products featuring on-demand capabilities in its most promising market segments: Web services, open standards, grid computing and self-healing systems.
Yet IBM is not alone in its attempts to expedite the concept and its technologies. Hewlett-Packard, Sun Microsystems and others are also incorporating their versions of on-demand into their own products.
Big Blue has invested millions to deliver on-demand computing -- and much of that investment has gone toward marketing and sales, education, acquisitions, research and development, hosting facilities and on-demand design centers where customers can test the new concepts.
Analysts say that the on-demand concept is likely to be the standard computing environment by 2007.
But IBM's greatest challenge, says Tom Kucharvy, president of Boston-based Summit Strategies, may be having to convince customers to take a gamble now in the hope that the payout will be huge when the vision pans out.
Kucharvy, who just published a report on IBM's on-demand computing initiative entitled, "Will IBM's On-demand Initiative Leave Competitors in the Dust?" says on-demand may sound like a lot of hype now, but it will happen.
"It's a slam dunk," he said. "The returns are so obvious."
Every major vendor and system integrator believe the technology is going to facilitate more effective utilization, he added. "The only thing that can change that is total failure of customer adoption."
It's unlikely that would happen, say analysts, but there is a lot of confusion surrounding on-demand, and some users aren't exactly itching to adopt it.
"IBM has a lot of education to do," Kucharvy said. "It's a really complex set of considerations that are involved here. The problem is different people within IBM still have different perspectives. They have the same message, but when they drill down, the perspective changes depending on where they're coming from. This is complicated, and it's hard to get across a cohesive message."
Leap of faith
"I don't think that on-demand computing is a bad idea," said Tim Granatir, vice president of technical services at Interlink Technologies in Maumee, Ohio. "But I'm skeptical about IBM's motives and especially their eventual pricing structure.
"In a lot of ways, I feel like IBM [is selling us] a bill of goods," he said.
Granatir cites when IBM had the idea to separate batch and interactive CPW on the iSeries and then charge a hefty premium for the interactive feature. According to Granatir, IBM put an artificial governor on the processor that essentially brought the system to its knees if the interactive threshold exceeded capacity. Your choices at that point are to rewrite your application to use less interactive processing power or shell out big bucks to IBM to let your processor do more interactive work.
"A lot of people have been burned by this," Granatir said, "and IBM's processing on-demand resembles that model in that the processors are already in your machine, but you pay IBM to allow them to be activated."
Kucharvy said he understands the reluctance to adopt the concept, and he's not surprised by the skepticism.
"IBM hasn't done a good job of explaining it," he said. "Vendors are using different terms and it's hard to compare. Some technologies are ready and some are not. It's hard to judge whether a vendor is really able to deliver on this."
He said the reluctance also stems from the fact that deployment of on-demand could result in job loss.
"That's why the top [upper management] says yes and the bottom [programmers] says wait," he said. "The top gets carried away in the concept and the huge potential it has to cut costs. It's really, really attractive. To the lower level, it's their jobs. The CIO isn't worried about his job deploying this."
Gordon Haff, senior analyst and IT advisor for Nashua, N.H.-based Illuminata, said that to say IBM is forcing users to take a leap of faith with on-demand isn't entirely accurate, but he too understands the hesitation to buy into it.
"On-demand is not really a technology or a product," he said. "It is really more of an idea and certainly while technologies exist today making it real, it's a broad concept."
Certainly, IBM isn't expecting businesses to throw out their current technology and implement on-demand, said Kucharvy. However, there's a lot of good technology out there featuring on-demand capabilities that can be implemented now.
But he warns that there will lots of steps along the way -- for better or for worse.
"Some implementation may not turn out that well," he said. "Everything isn't fully baked. Second, to get the full benefit, you've got to change processes and jobs definition and skills, and some companies aren't going to be able to adjust as well as others."
Kucharvy said that acceptance of on-demand is going to be a gradual process in which businesses start with a specific implementation, likely their greatest pain points, where they can get the best and fastest return on investments. Then businesses will be able to evaluate each project on its merits.
However, businesses put themselves at risk by implementing products with the idea that these products will ultimately lead to the dynamic on-demand environment that the vendor promises. Businesses should implement the technology to solve very specific problems, not as part of an overreaching long-term on-demand plan. The idea is to do what you've got to do to solve current problems, don't get caught up in a vendor's vision that has no guarantees.
"Yes, there are going to be some disappointments," Kucharvy said. "But taking an incremental approach is a no-lose proposition if they're well researched, well thought out."
FOR MORE INFORMATIONIBM buoys on-demand with new iSeries, payment plan IBM pumps up on-demand line with new iSeries model Finding utilitarian uses for utility computing