"I know that you and Frank were planning to disconnect me, and I'm afraid that's something I cannot allow to happen," HAL 9000, the fictional mission computer, said in "2001: A Space Odyssey." Stanley Kubrick's film may date back more than 40 years, to 1968, but the monolithic intelligent mainframes that IBM first produced in the 1950s are still a mainstay for financial firms -- and fear of disconnecting them may be one of the primary reasons, according to Aite Group senior analyst David O'Connell.
"Mainframes are integrated with so many business processes that no one really wants to touch them," O'Connell explains. There are a lot of devices these days that can do what mainframes do, he acknowledges. But there are very few reasons to migrate off of the mainframe and many reasons not to, he adds.
While mainframes' functions are fairly well documented, like HAL, there is still an air of mystery about them, at least for non-engineers, who are typically reticent to simply switch them to the "off" position. "You never know if it might switch off a trading desk --no one knows exactly how integrated they are," O'Connell relates. "No CIO wants to say, 'Gosh, the reason we have an outage at an entire retail network is because we shut down the mainframe to save $30,000.'"
Fear of Failure
Of course, fear of the consequences of turning off these machines is not the only reason behind their continued popularity. They actually still offer value. Mainframes support huge portions of the data environment at most large enterprises. And as firms search for new ways to sift through and analyze mountains of structured and unstructured data, mainframes still offer one of the best ways of doing this, suggests Ken Muckenhaupt, CTO for the financial services sector, IBM Systems & Technology Group (hardware division). "For all intents and purposes, the financial world runs on this class of computing," Muckenhaupt says, noting that mainframes are used today for traditional and online transaction processing, including generation of reports, account management and core banking operations.
"Firms say, 'I need to have more capacity. We're being swamped,'" Muckenhaupt continues. "Nothing beats the mainframe if the client needs extreme reliability, availability and serviceability. It's second to none. You'd need to go through many more hoops or configurations to come close to what a mainframe would do. It has unlimited scalability in virtual environments, which puts it far above distributed environments."
One area in which mainframes play a critical role is analytics, where mainframes serve as a primary data source for big data capabilities and solutions such as Hadoop, according to Aite's O'Connell. Since these behemoths can easily crunch data in real time relating to the history of account holders and their behavior, mainframes are used to analyze opportunities to create competitive products as well as for fraud mitigation, he adds. Meanwhile, new regulatory reporting requirements generated by Basel III and Dodd-Frank are dialing up the pressure on firms to mine data in near real time, increasing the reliance on mainframes further.
Ameriprise Financial CIO Randy Kupper says his organization, which just signed a multiyear partnership agreement with IBM to incorporate the vendor's technology in its data centers, uses the mainframe for interactions with outside wholesalers and other manufacturers of products that its advisers sell. The mainframe's analytics capabilities are a huge draw. "There's a lot of batch processing and data crunching," he indicates, noting that Ameriprise works with research firm and investment manager Morningstar, a provider of data on more than 385,000 investment offerings, along with real-time global market data on more than 8 million equities, indexes, futures, options, commodities and precious metals.