) focused on giving the mainframe an expanded role in the technology plans of large corporations. Put simply, the goal is to make Big Iron once again the hub in corporate computing.
The IBM plan is to extend two of the key attributes of the mainframe -- security and manageability -- to the entire datacenter. That means the technology will increasingly be applied to any kind of machine in the datacenter -- no matter who sold it.
"We're extending our mainframe capabilities to everything that's connected to the mainframe," says Erich Clementi, general manager of IBM's zSeries business unit. "Your transaction goes through the Cisco (CSCO
) router to the Unix server to the mainframe, and it's all seen as one business process."
WAY OF THE DINOSAUR. Analysts reacted positively to the new strategy. One reason: The proliferation of machines and technologies has corporate techies searching for ways to simplify their jobs and protect their data from prying eyes. "I think this will generate a lot of interest, and not just from the longtime mainframe fans but from all sorts of people who are looking for solutions to the problems in computing," says analyst Mike Kahn of researcher The Clipper Group.
IBM is investing heavily to make the mainframe more relevant. Over three years, the company spent $1.2 billion and enlisted the help of 5,000 engineers. The new machine also delivers double the processing power of its best-selling predecessor, T-Rex, introduced two years ago -- which kicked off a six-quarter growth spurt for the long-sluggish mainframe.
Will z9, which had been code-named Danu, prompt another huge wave of buying? Not clear. IBM's mainframe sales have dropped for three quarters in a row. The peak was the fourth quarter of 2003, when sales hit $1.66 billion. The low was the first quarter of this year when IBM pulled in $939 million in revenue. In the last three quarters, IBM's mainframe revenue has declined 4%, 16%, and 24%, respectively.
TRADE-OFF. Declines are typical when one product cycle is ending and another one is beginning. But IBM's rivals insist that something more fundamental is going on. "You can't defend a revenue decline of that size by saying, 'Our customers are holding out for the next product launch,'" says Rich Marcello, a senior vice-president at Hewlett-Packard (HPQ
). "It says IBM customers are finally realizing they've been paying too much for systems that no longer have a stronghold on the market." IBM doesn't disclose mainframe prices but customers say the machine starts at $1 million and goes up to several million.
Analysts, too, believe the T-Rex phenomenon will be hard to match. Z9's performance gains surprised some, who were expecting something less dramatic. Still, they say the T-Rex boom was caused in part by a lot of pent-up demand at the time, combined with a huge performance improvement and a 25% reduction in the price of computing power.
Several mainframe customers interviewed by BW Online don't anticipate a buying binge, either. Edward Ciesla, assistant vice-president for technology planning at Bank of New York, says the bank plans on buying two large Danu mainframes for its new data center in Tennessee, but will retire four smaller ones at the same time. The outfit still plans on running other applications, including its brokerage business, on Unix machines. "The new machines give us a lot more processing, but we're not adding that many applications," he says.
BEYOND THE BOX. IBM's challenge is to convince many of the 18,000 companies that own mainframes to use them in new ways. It has made some headway. Ken Kucera, the chief information officer at First National Bank of Omaha, retired 30 Unix servers and moved all of the applications to a single mainframe, completing the job this spring.
Kucera is among a cadre of venturesome corporate tech leaders, many of them in banking, who are using the mainframe as a sort of computing factory, often using the open-source Linux operating system or the Java programming language to modernize the machines. "My goal in life is to simplify my datacenter, not make it more complex. My strategy is back to the mainframe," says Kucera, who expressed interest in using it as a hub for corporate computing.
It was customers like Kucera that convinced IBM to extend its mainframe capabilities beyond the box. IBM's eBusiness Council of corporate users began asking the company two years ago to extend its mainframe-encryption technology so they could protect their data after it leaves their property on the way to tape-storage warehouses, according to George Walsh, IBM's vice-president for the systems environment.
EDUCATION INITIATIVE. IBM then came up with the idea of allowing customers to extend management of encryption to additional computing and data-storage systems. Also, with z9, customers can use their mainframe to optimize the flow of data within a network of servers. "Right now, datacenters are like lose threads. This stuff is the Kevlar that holds things together," says Walsh.
One of IBM's problems is making sure there are enough programmers and technicians with mainframe knowledge to keep the technology going. Market researcher Meta Group three years ago predicted a shortage of people with mainframe skills by 2007. IBM responded by launching a program that provided colleges, now numbering 150 worldwide, with grants and study materials to help them train people with mainframe skills. So far, several thousand students have received the training. The goal is to train 20,000 by 2010.
There's a lot at stake here. If IBM can convince enough customers that its new "hub" technology is worthwhile, the positive momentum started by T-Rex might continue with z9 -- and the mainframe could once again be a hot technology. If not, it will likely play an ever diminishing role in the computing universe. Hamm is a senior writer forBusinessWeek in New York