Sales Leap Defies Anti-Mainframe Campaign
After a quarter like System z just had, it’s difficult to see how or why competitors are still writing epitaphs for Big Iron
Late last month, Hewlett-Packard Co., Intel Corp., and Oracle Corp. took the wraps off of a new partnership to get existing mainframe customers up and running on Intel-based HP servers. The announcement came on the heels of an unprecedented arrangement between IBM and Oracle designed to help encourage Linux-on-zSeries adoption.
More to the point, it came even as IBM officials were touting one of the best quarters on record for Big Iron, with 18 percent year-to-year growth in MIPS shipments and revenue growth clocking in at seven percent per quarter.
In other words, while the HP-Intel-Oracle partnership hopes to capitalize on widespread disaffection with the mainframe, the opposite sentiment—if buying patterns are any indication—seems to be the prevailing sentiment.
True, mainframe sales did dip in Q2 this year, but that was also ahead of IBM’s z9 Business Class (BC) and z9 Enterprise Class (EC) system refresh, says Bob Hoey, worldwide vice-president of System z sales for IBM.
Taking all of this into account, IBM officials claim, it’s difficult to see how or why competitors are still writing epitaphs for Big Iron. The irony, in fact, is that officials like Hoey and System z GM Jim Stallings have spent much of 2006 doing the opposite—in other words, writing epitaphs for, or predicting a slump in the popularity of, distributed Unix, Windows, and Linux systems.
After the quarter System z just had, it’s not difficult to see why. “There was a 25 percent year-to-year revenue growth in mainframe [sales], and we had 18 percent year-to-year MIPS growth. We had double digit revenue growth in all of the [regions] around the world, and our market share increased by five percentage points in the greater-than-$250,000 server category,” said Bob Hoey, worldwide vice-president of System z sales for IBM, last month. “This was the highest third quarter MIPS shipment in the history of this product, and we believe part of the reason we were so successful has to do with the fact that customers are really responding to [the mainframe] value proposition.”
There’s a further wrinkle here, too. The mainframe hasn’t traditionally been perceived as an SMB-friendly option. That started to change half a decade ago, when IBM introduced its first z800 (“Baby Z”) mainframe systems, which it marketed as Linux-only machines designed to support non-traditional mainframe workloads. Since then, Big Blue has shipped a dedicated small- and medium-sized enterprise system (the z890) as well as the z9 BC. Clearly, analysts say, Big Blue is thinking outside the proverbial cabinet when it comes to ginning up new mainframe sales—and helping vouchsafe the future of that platform, too.
Big Blue drove home this point earlier this year at its Mainframe Software Summit when it published TCO results which purport to demonstrate that IBMmainframes have a lower overall TCO than Unix, Linux, or Windows platforms.
How is this possible? For one thing, says analyst Wayne Kernochan, formerly of InfoStructure Associates and now with Illluminata Inc., mainframe licensing costs—the historical thorn in the side of mainframe shops everywhere—are now actually competitive with other platforms. “A key reason for this is IBM’s focus on driving these costs down. For example, IBM’s database-administration tools have had the effect of driving the prices of competitors such as CA and Compuware down significantly,” Kernochan writes, adding that IBM licensing costs typically decrease per unit of workload as workloads increase.
Moreover, Kernochan points out, Big Blue’s specialty processor engines—e.g., Integrated Facility for Linux (IFL), zSeries Application Assist Processor (zAAP), and zSeries Integrated Information Processor (zIIP)—have been instrumental in reducing mainframe TCO for many hard-hit customers. While there’s some debate about the value of these processor engines relative to the performance of the same kinds of workloads running on z/OS, no one disputes that for customers who choose to deploy Linux, WebSphere, or DB2 workloads on the mainframe, Big Blue’s specialty processor engines are nothing less than a godsend.
Kernochan also singles out the mainframe’s integrated power and cooling management capabilities, its compact (all-in-one) form-factor, and—of course—it’s ever-improving scalability. All told, he concludes, IBM can credibly claim that the mainframe’s TCO advantages should increase over time, helped in part by aggressive pricing that has helped reduce costs by as much as 50 percent over the last four years.
Speaking on the heels of System z’s record Q3, IBM officials were buoyant. Hoey, for example, cited worldwide demand for mainframe systems as one especially important revenue driver. Indeed, while growth in mature markets (such as North America and the EU) remains steady, emerging markets—and especially Euro-Asia and Asia-Pacific—could become hotbeds of growth.
“When we announced the z9 Business Class, the low-end machine, earlier this year, instead of having the product announcement in New York City, where we usually have mainframe announcements, we decided to make the announcement over in China. The reason we did that in a symbolic way is because China has about eight million SMBs, and that number is growing about 10 percent per year,” Hoey commented.
“We believe we can sell the low-end mainframe to [SMB} customers who do not have a mainframe today. We believe that if we address firms who have some number of Unix or Intel servers—say, 25, 35, or 45—we believe we can sell them, start them on a mainframe by replacing them with a single engine mainframe, running an IFL Linux engine.”
That’s why IBM’s recent $100 million push to simplify mainframe administration and software development is so important, Hoey says.
“What we’d have to be able to do is convince those customers that are running those distributed servers that provided they understand how to do the systems administration … with a relatively small investment of time (say, six months), we can take someone who’s already computer-literate and provide them sufficient programming and administration skills in order to move them to a mainframe environment,” he concludes.