Suffering from a case of data overload? Sick of poor performance from distributed computing systems? If so, IBM’s System z mainframes may be the cure for your malaise—as long as you can negotiate the right price for the treatment.
It’s been years since mainframes were in their heyday. But they remain entrenched in many corporations, like ExxonMobil, which has used the technology for almost 40 years for 3D seismic imaging.
Mainframe systems are designed to handle massive amounts of data processing functions centrally; the information typically was accessed from “dumb” terminals that simply displayed what the mainframe returned. This architecture, however, fell from favor as too expensive and unwieldy. Instead, the dominant application model became client/server, in which processing functions are split between cheaper server systems on the back end and a user’s own computer.
But now, some companies are making their way back to mainframe big iron, or they’re expanding their use of it. And Big Blue, customers say, makes the best mainframe hardware on the market. To many, its System z mainframes, the new brand name for what was previously called the zSeries, provides unmatched reliability, high-capacity performance and security.
While pricey—an entry-level z9 system, for example, starts at $100,000—the savings can make a mainframe worth the investment. Take the experience of Mark Shackelford, director of information services with Baldor Electric. The maker of electrical industrial parts used an IBM z990 to consolidate SAP applications previously running on 10 Windows servers and six Unix servers. Shackelford says that helped cut his costs by 50%. In large part, that’s because just three people are required to manage the mainframe; it took 20 to oversee his old systems. Plus, his 6,000-square-foot computing room is now “basically empty.”
But some customers have felt pinched by the high list prices for mainframe software. In 2002, Blue Cross Blue Shield of Tennessee wanted to offer its 2 million members new information on its Web site, including applications to find doctors and hospitals and look up claims information.
Bob Venable, manager of enterprise systems at the health insurance provider, struggled with rising mainframe software costs. The company was using multiple mainframe applications (to handle tasks like monitoring system performance) from several vendors, including BMC and CA.
So, Venable called in IBM to figure out the most cost-effective programs to run on its zSeries servers. The result: He says the company saved more than $14 million in data-processing costs in less than two years, by consolidating all its mainframe programs on IBM’s versions. “The real savings on the table these days are in the software,” he says. “It’s picking your software to minimize the costs of your hardware.”
Marq Youngblood, CIO of the Oklahoma Department of Human Services, ran into a similar software-pricing wall. In 2002, a key child-welfare database running on a Hewlett-Packard Unix platform faltered. Youngblood and his team decided to move the application to an IBM mainframe platform running Linux.
But IBM and third-party mainframe programs “were too costly,” Youngblood says. After a lengthy evaluation and negotiation process, Youngblood and his team decided to use 25 of IBM’s mainframe applications. (He would not disclose details of the deal.) “We addressed [cost] and got to a reasonable point,” he says. “That’s why we’re still on the mainframe.”
What’s more, newer IBM mainframes are less expensive relative to the processing power they deliver. That’s key for Big Blue, as customers consider lower-priced, high-performance servers available from competitors and IBM itself. Between 2004 and 2005, IBM’s mainframe system revenue dropped 7.6%, according to company reports. At the same time, the volume of mainframe hardware IBM sold—as measured in million instructions per second, or MIPS, a standard unit of mainframe computing power—grew 7%.
Asked to explain this dynamic, an IBM spokesman says that newer specialty mainframe processors (such as those designed to run Linux) have “altered the cycle of sales” for new systems. In other words, customers should be getting more bang for their buck. According to IBM, the System z9 109, introduced last summer, has a 40% better price/performance ratio (the cost of a system divided by processing capacity) than the z990, its previous highest-end model.
Still, more attractive pricing isn’t everything. Youngblood notes that his team hit a few snags. For instance, one of the IBM software products (which he wouldn’t identify) didn’t work on the new platform, despite IBM’s promises. When asked for a response, IBM would not comment.
Youngblood says the disruption was minor—his team got the software in question running with help from IBM—but points out that due diligence is critical for any big project. “You don’t take a vendor’s word,” he says, “not even IBM’s.”
|
IBM Financials Results*
|
* For fiscal years ended Dec. 31
** Includes servers, storage systems, microelectronics, engineering and technology services, retail store solutions and printer systems
† Includes systems and technology group, plus personal computer division, which IBM sold to Lenovo in April 2005
Source: IBM Regulatory Filings
Story Guide:
Straight shooter: CEO Rex Tillerson Doesn’t Play Games
- Big Oil’s Big Challenge
- R&D Like Launching a ‘Moon Shot’ Every Decade
- Exxon’s Balanced Methodology
- Virtual Drilling: Exxon’s Technology Edge
- Business Intelligence Guides Exploitation of Oil Resources
- Oil in the DNA: The Exxon Culture
- ExxonMobil Base Case
- The Technology of Oil Exploration
Context:
- How Big Oil Uses XML
- Roadblock: Getting Oil Business Managers on Board
- Gotcha: Don’t Let Grid Computing Trip You
- Vendor Profile: A Look at IBM’s Supercomputers and the Oil Business