Channel Insider content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

Oracle Database 10g be damned—IBM has been doing grid for longer than Oracle, better than Oracle and for cheaper than Oracle, its database executives want us to know.

New features in the recently released DB2 Universal Database 8.2 deliver “a very simple way to set up a high-availability failover environment with a much more cost-effective value proposition than our closest competitor: up to 60 percent savings in some cases,” Bob Picciano, IBM vice president of database technology, said in a recent interview with eWEEK.com.

But is IBM capturing the hearts and minds of grid customers outside of academic institutions and huge research and development labs, or is Oracle’s 10g message gaining ground?

Picciano says yes, IBM most certainly can compete, pointing to proof in the form of the recently released 2004 rankings by Waters magazine, a financial publication whose readers selected IBM as their No. 1 choice for a grid management provider.

IBM snagged 48.4 percent of the vote; that’s double the ranking of second-place Oracle Corp., which was selected by 22.2 percent of the 423 respondents.

“While many corporate technology vendors have yet to incorporate grid technology into their solutions, IBM has already created offerings for a number of vertical industries, including the financial services industry,” Waters editors wrote.

Some agree. Dan Kaberon, director of computer resource management at Hewitt Associates LLC, refers to the two projects he’s running on IBM and DataSynapse Inc. grid technology as “the 2003 miracle” and “the 2004 miracle.”

Hewitt, which is a global human resources outsourcing and consulting firm that delivers human capital management services including payroll, is currently running two applications on grid. Both were typical grid candidates: compute-intensive programs that sucked up a lot of mainframe horsepower and that were, consequently, expensive to run.

In 2003, Hewitt tackled a mainframe application that calculated pension benefits. Hewitt services benefits administration for large employers. External news events cause great spikes in demand for data—think of mergers and acquisitions, wars and natural disasters and how they influence people’s decisions to check their pension benefits or to request information on when they might retire. Because of such unpredictable flux in resource demand, planning horsepower was very difficult, Kaberon said.

Hewitt converted the pension application to run on Intel Corp. blade servers running Linux. With a “tremendous amount of help” from IBM’s Center for Transaction Processing, Hewitt hooked up the blade servers to run in conjunction with the mainframe, thus offloading compute-intensive work from the expensive-to-run mainframe and onto the cheaper servers.

Next Page: Grid miracles.

The 2004 miracle concerned a third-party application that generates what Kaberon refers to as composed print. Composed print comprises Postscript and PDF files that do things such as define contributions and quarterly statements.

During open enrollment periods in the fall, Hewitt would generate five- to 10-page documents tailored to each of its millions of participants. That process was redefined so that the DB2 databases could still live on the mainframe, but the work itself was parallelized—parceled out as chunks for the blades to chew through and feed back into the mainframe for recomposition.

Without grid, one print composition job alone could have taken two weeks to run, Kaberon said. With grid, it took 14 hours—a time savings that represents reduced costs of 90 percent or more.

If Kaberon’s experience is indicative, it’s easy to see why IBM is scoring so high with financial services customers for grid management. He pointed to ugly code-conversion snags that arose in the two projects, such as the disconnect that arose from the mainframe doing all of arithmetic in PAC decimal—a numeric representation that doesn’t exist in an Intel machine.

“Fortunately, IBM rode in on a white horse and provided the translation,” Kaberon said. “They did it all for us.”

IBM, with its long-term relationship with grid technology company DataSynapse, has many such grid glory stories to which it can point. And it’s certainly on target to rule the roost in more financial institutions. Back in January 2003, it launched an initiative to fine-tune its grid technology for a handful of verticals: aerospace, automotive, financial markets, life sciences and government.

To read more about grid computing’s advances on Wall Street, click here.

According to Pat Aughavin, vice president of business development at New York-based DataSynapse, IBM and DataSynapse are a good fit for financial services companies because they target things near and dear to the vertical industry’s heart: security, scalability and reliability. Those essential ingredients are not necessarily so important in the world of academic grid computing, nor do they garner the attention of generic grid vendors, she said.

“We grew up here in the financial services market and understand security, scalability and reliability aspects of processing their applications,” Aughavin said. “That’s inherent in our products, which you don’t find in many other generic grid solutions developed for the academic marketplace, where a lot of compute power was thrown at a problem, but if something broke, it was just restarted. There weren’t the time constraints as in the financial market.”

Does Oracle understand such industry-specific nuances? Can Oracle steal away the attention IBM garners in this important vertical?

Next Page: There’s no comparing Oracle with IBM.

Some say the question is, quite simply, a non sequitur, like comparing apples to oranges. “What Oracle calls grid is rather different than what IBM calls grid,” said Frank Gillett, a Forrester Research Inc. analyst.

“Oracle’s talking about machines or nodes, whereas much of IBM’s usage has little to do with databases and more to do with what DataSynapse does. The financial services community is using … a different definition of grid than what Oracle’s talking about when they’re talking about grid.”

At the heart of the dilemma, Gillett said, is that the industry is now host to a medley of terms defining the hazy notion of shared resources: utility computing, on-demand, N-1, adaptive.

“What Oracle is on is an interesting track,” said Gillett, in Cambridge, Mass. “What they’re doing is making their equipment more sophisticated and automated, but … we believe it’s a subset of the grander vision, which has no clean name in the marketplace but gets called N-1 or adaptive or on-demand.”

Click here to read about Sun’s vision for grid computing.

What does the murkiness of grid’s definition do to IBM’s desired title as king of grid? It makes it impossible to defend or to challenge, since the company’s grid offerings differ so greatly from those of its biggest rival, Oracle, experts said, and it underscores the fact that rankings should be taken with a grain of salt.

Oracle couldn’t provide comment for this article by the time it was posted, although a spokesman agreed that the current terminology around grid computing is confusing. A follow-up article will examine Oracle’s strategy around the financial services market and Oracle’s success in getting its grid message across, now that the first anniversary of 10g’s release is close at hand.

Check out eWEEK.com’s for the latest database news, reviews and analysis.