HP In The Glass House?

Move over MVS. HP-UX is moving in.

"We're making the HP 9000 environment look like MVS...without all the problemsthat come [with it]. It's the second attack on the mainframe."

-- Nick Earle, vice president and marketing manager for HP's EnterpriseComputing Solutions Organization

That statement, made by a feisty Nick Earle, during an interview at last year's HPWorld in San Diego (see HP Professional November 1998), positions HP for battle in theglass house. But a larger questions remains: Is there really a war for data centersupremacy being waged by HP and IBM in the hearts and minds of mainframe loyalists or isit more like detente for UNIX vendors?

"The diehards who have remained on mainframes are those who find that mainframestruly have some robust bulletproof characteristics that UNIX systems can't reallyduplicate," says Rich Partridge, senior research analyst for platform andperformance, at market research firm D.H. Brown Associates Inc. (Port Chester, N.Y.).

While there may be some fight-to-the-last-mainframe outposts still out there, theco-existence trend may be inescapable. "Microsoft and Intel are trying to pushIntel-based servers. UNIX vendors are trying to push UNIX to people who have gotten awayfrom the thinking of [only having] mainframes in the data center. And NT's not an option;so, of course that leaves UNIX," says Joyce Becknell, director of UNIX marketing forThe Aberdeen Group (Boston, Mass.).

In the past, UNIX vendors, and HP in particular, have described themselves as mainframealternatives. "While this is an attack on the data center, I believe that [UNIX] willbe more successful at surrounding mainframes in offloading some functions like decisionsupport and data warehousing rather than replacing mainframes that are doing onlineproduction, mission-critical work," says Partridge. But he does give HP some creditfor taking enterprise functions and putting them on [HP] platform. However, he says,"HP is not the death knell for mainframes. While HP marketing slogans get attentionby saying mainframes are dead, the real opportunity is to do the new growing applicationslike decision support and data mining ... that could be put on systems of the Vperformance class."


And that may not be a bad thing, he adds, considering the acreage of business LAN-scapeto be won. "In reality, [HP is] going to find a lot of business at the high end.Clearly, that's what Sun has done with the [Enterprise] 10000. They had a few slogansright away when they launched it with full-page ads about the frontal attack on themainframe." That tactic of frontal assault quickly became one of attrition as they"... maneuvered themselves with the 10000 into a position that it's not displacingthe mainframe, as much as absorbing new growth in the enterprise."

"One of the problems ongoing is that there are so few truly enterprise [UNIX]applications, it's still very difficult to unite a system to get full strength," saysBecknell. "Users are always looking for more power. So you have all the UNIX vendorstripping over each other trying to figure out how to win customer loyalty and build theirproduct set."

As to the combatants in the war to break through the walls of the glass house, it stillboils down to HP, Sun and IBM -- with no clear-cut leader. As far as HP is concerned,Becknell is cautious. "They've got a good name. And they're trusted by IS and peopleare happy with them. So that's the strength they can leverage into the high-end datacenter," she says. But HP has never been a strong software player. [They've] alwayshad the most tenuous relationships with software partners."

The old adage about users buying the software first then selecting the hardware tomatch is true, Becknell says. For that reason, while HP should continue to score big withits traditional customer base, without stronger ISV relationships, winning new businessmay be tough. "You really need to connect with [ISVs] and provide a total solutionand you need to be thoroughly in bed with them," she says. "I'm always going tolook to HP to show me how solid those relationships really are."

Becknell sees the V2500 and the related enhancements as important weapons for HP goingforward. "The [V2500] preliminary numbers I've seen are certainly impressive. It'sdefinitely fast," she says. "And, they're also working to bring in the[enterprise] services like ContinentalCluster, disaster recovery and backup. And ofcourse, they've really committed to the five 9's strategy," of 99.999% percentuptime.

While performance numbers and benchmark tests aren't everything in a competitivemarket, customers still look for a certain minimum and a certain level of capability."IBM wins on their services and they win because people already think of them as adata center [vendor]. The RS/6000 is popular and they have the System 390," she says."However, the problem with the RS/6000 is, you can't name the last time it won abenchmark. So, if you're [comparing the] RS/6000 versus the HP 9000 or the Enterprise10000 from Sun, then the RS/6000 becomes a little bit shakier."

While benchmark performance may be the bullet that chips some glass, it's enterpriseapplications and services that will shatter the glass house door. While IBM wins the datacenter name-recognition factor, HP's loyalty to both its users and platforms are importantpoints in its favor. "You're seeing [HP] make a commitment to their customers andmake a commitment to their products. I think HP has made very clear that they're committedto HP-UX and PA-RISC at least through 2002 or 2003, and of course beyond that is anybody'sguess for any processor," says Becknell. Still she thinks, "the only supplier ofsystems who's good with software is IBM. None of the rest are software companies. And, HPhas good service but it's nowhere as good as IBM."

But in her mind at least, there remains a litany of ubiquitous questions about IBM:"I'm not sure I understand IBM's commitment. What they're doing with the processorand what they're doing with AIX and why they haven't committed AIX to Merced and nowthey're going to rewrite it with SCO."

Partridge generally agrees with that view, but raises Compaq as an off-in-the-distancebut gaining competitor. "Competitors tried to paint Compaq as buying Digital for theservices arm and then kill off the rest of the business," he says. "Clearly,they intend to continue to work on Digital UNIX and the Alpha server."

As the industry accelerated its UNIX development cycle pace, Digital lagged behindbecause engineers kept looking over their shoulders to see where the axe was going to fallnext with layoffs. But, now emerging from what Partridge called the "quiet time"since the take-over, he predicts, "Compaq will fund Digital's catch-up and Compaqbought them because they want to be in the high-end."