RealTime IT News

On-Demand Computing: From Fuzzy to Focus

Ask executives at NIBCO, a manufacturer of plumbing devices, about what the tech term "on demand" means to them, and they will point to China.

When Chinese manufacturers can produce the same flow-control valves and devices as the 101-year-old business -- at half the cost -- where does NIBCO build its profits?

"We compete in an international market in a commodity business, where we have small margins on products," said John Hall, director of supply chain systems for the Elkhart, Indiana-based NIBCO.

That's the cold reality in commodities.

"Whether it's copper fittings, plastic fittings, or an inch-and-a-half plastic fitting that carries water from a house, it's the same fitting -- no matter who makes it," he added. "What [Chinese manufacturers of the same pipe fittings] can't provide is quick answers to the customer and immediate service."

NIBCO realized that, with 12 manufacturing plants and distribution centers, it needed to draw out accurate inventory targets from its sales data in order to fill last-minute change orders.

With the help of IBM's business consulting division, NIBCO re-architected its SAP database in order to improve the granularity of sales feedback and to build better turnaround data. The three-month project focused on building customized inventory-optimization algorithms that helped NIBCO streamline its inventory management.

The new architecture integrates the company's systems across all 12 manufacturing plants. Using weighted average daily sales, as well as a host of new algorithms based on data from three main zones in which the company operates, NIBCO is able to increase its lead time by knowing how many times a year it needs to make its products.

Now, when a customer needs a particular pipe fitting, NIBCO can pluck it out of what used to be a needle in a haystack of production sites.

"We delivered somewhere around a 13 percent to 14 percent reduction in inventory," said Hall. In a company with about $500 million in sales, the system is saving millions already, thanks to better asset utilization.

"We built an architecture that supports key initiatives that business folks needed," said Tim Ronan, NIBCO's business systems analyst, "not only by using our historical sales demand, but with our reporting capability, and our product alignment. This allows us to quickly alter production in order to match changes or potential changes," he said.

In addition, the new system placed approximately 15,000 products online while automating the labor-intensive process. Just in time manufacturing? NIBCO would call this its on-demand computing strategy.

By Any Other Name

Computing services on-demand, by-the-drink, as you need it, utility- style. The term can be difficult to nail down. Charles King, senior IT research analyst with Pund-IT Research, said an easy way to sum up the term is to call it "computing when you want it, the way you want it."

Industry examples abound -- in many different flavors.

Google is a kind of on-demand provider. It provides you with search results when you bring up the site. Google's recent splashy deal with Sun Microsystems to provide some sort of office productivity functions over the Web is certainly a software-on-demand offering -- once it materializes.

Microsoft , aware of the threat of the shift to software-on-demand in a world where which operating system one is running doesn't matter so much, recently announced a reorganization of its business divisions in order to make its software more adaptable for this trend.

Arena Solutions offers on-demand product lifecycle management software that's delivered only over the Internet: software-as-service, subscription-style. Like Salesforce.com , which provides customer relationship management software to customers via a Web portal, Arena is part of a diverse yet growing group of technology vendors offering "computing your way."

Systems vendors have their own names for it. One version is HP's adaptive enterprise. IBM's version is about data center services as a utility and strategic on-demand projects such as NIBCO's. Sun's is called SunOne.

For the most part, tech research firms talk about this trend as deploying software and middleware integration, open standards, virtualization and autonomic (or self-healing) computing systems in order to deliver a "real-time enterprise" for customers.

The Time Is Ripe

The holiday online shopping rush may give another boost to the concept. Pund-IT's King said as the holidays approach, more online retailers will deploy on-demand payment systems as they order up extra computing capacity to handle the holiday rush. "In the old pre-capacity upgrade model, you basically bought three to four times as much capacity as you needed, and most of it sat idle until it needed to come online at high demand time," he said.

"The capacity upgrade of on-demand can either be set up to provide additional processor capacity manually, or it can actually be set up as a policy thing so you're basically paying for capacity."

IBM On-Demand

But King said IBM is considered to be the most committed to its strategy right now, because of how it is integrating its approach across all of its divisions.

"Utility processing is only a small part of the story" of on-demand," said John Lutz, vice president of IBM Worldwide's On-Demand Business division. "To just call on-demand 'utility computing' amounts to a narrow view of the world," he told internetnews.com.

Initiatives range from providing autonomic healing systems in its server hardware, deploying its consultants to help build advanced database applications like NIBCO's, and even innovative financing from IBM's Global Financing division, which is part of IBM Global Services.

With $35 billion in assets to put to the financing cause, IBM ranks as the largest IT financier on the planet, said Paul Foulkes, worldwide vice president for project financing for IBM Global Financing.

As the division of IBM Global Services division that's gone from $10 billion in annual revenue in 1994 to close to $50 billion in 2004, the financing group is out in front of Big Blue's on-demand strategy; it offers a competitive advantage by helping clients take the plunge on major projects. Where one client might have been leery of a business process transformation project that entailed a soup-to-nuts reengineering of the company's network for Web services deployment, IBM's financial division eases up on the capital outlay and lets the customer pay it back as the project starts to show a return on investment.

As recently as a few months ago, Lutz said, he was seeing a shift in the conversation about on-demand, as more customers and vendors look beyond the idea of squeezing more dollars out of their CPU investments.

While IBM's iSeries and Blade Center server platforms provide scalability and can be used as points of management for widely dispersed server environments, King said it's not about the labels on the hardware. It's about computing when you want it, not how the vendor thinks you want it.

He said, "It's a destination with a thousand different roads that lead to it."