|
HP Eats Compaq, Preserves OpenVMS, and We Mull an OS/400 Union
by Timothy Prickett Morgan
The Delaware Chancery court judge hearing the suit brought by Walter Hewlett against the upper
management of Hewlett-Packard threw the case out of
court, allowing HP to complete its $19 billion acquisition of rival Compaq last week. HP has changed its symbol on the New York
Stock Exchange from HWP to HPQ, to reflect ; Compaq shares will no longer trade. Significantly for
OS/400 shops, HP has preserved its proprietary OpenVMS platform, and for good reasons that are similar
to IBM's continuing commitment to the iSeries.
The three most visible pieces of Compaq that remain after the merger are the "Q" that is now part of HP's
stock market symbol, the ProLiant server brand, and former Compaq CEO Michael Capellas, HP's new
president and chief operating officer. HP is most definitely calling the shots here, and upward of two-thirds
of the combined companies' middle managers are coming from HP's side of the house. Where Compaq had
market share advantages over HP in Intel-based servers
and corporate PC sales, Compaq executives and middle managers have apparently come out on top. This is
exactly what everyone expected.
HP's CEO and chairwoman, Carly Fiorina, gave a pep talk to press, analysts, and customers on Tuesday
afternoon, walking very briefly through the company's vision for itself going forward, its market position in
key IT sectors, and its product roadmaps. HP launched a revamped Web site that combines the offerings of
both HP and Compaq and archives the millions of pages that each company has online.
The net result of all of this, in terms of the Wintel and Lintel server market, is that ProLiant is now the brand
name of the product line, and that HP will roll NonStop, OpenVMS, and a unified Unix operating system--
comprised of the Tru64 clustering and file systems and the core HP-UX operating system--to future
Itanium processors from Intel. HP has been moving toward a future of Itanium-only server processors since
1996, and may actually get there by 2004 or 2005 after it is done digesting Compaq. And Compaq
announced early last year, before announcing a deal with HP, that it would dump the Alpha processor over
the long haul and move to Itanium. By late 2004 or early 2005, HP says it will have a unified Itanium
server line-up that spans up to 64 processors and can run Windows, Linux, the unified Unix, OpenVMS,
and the NonStop fault-tolerant operating system. We're gathering information on the detailed product
roadmaps right now, to give you a better sense of what HP is up to, and we will present this information as
soon as possible. HP, like rival IBM, which is roughly the same size as the new HP, believes that it will
need a mix of platforms to support its current and future customers. However, Capellas last week was
telling analysts that he expected Windows and Linux to "eviscerate" proprietary and Unix midrange
platforms--presumably including HP's own popular HP-UX minis--in the coming years, particularly as the
new HP pushes them very hard as the world's dominant server vendor.
Digital Equipment Corp. (the original minicomputer vendor from the 1970s, which created the VMS
environment for its VAX computers) and Compaq (after it bought Digital in 1998) have moved all of the
relevant and interesting Unix and related open systems standards into VMS, rechristening it OpenVMS.
IBM has made the same moves with its venerable MVS mainframe and OS/400 midrange environments,
and, as I have said before, MVS--now z/OS--could technically be branded as a Unix operating system
because it supports all the requisite APIs. OS/400 is missing only the APIs relating to supercomputer
applications; otherwise, it, too, could be branded as Unix.
How important this is to OS/400 or OpenVMS customers is something that customers have to judge for
themselves. It sure does make it easier to continue using these so-called proprietary legacy environments
and to make the case for them in the boardroom and the data center.
The main thing to remember is that HP--which sorely wants to ditch any platform it can--cannot ditch
OpenVMS. Digital and Compaq have collectively sold and supported somewhere between 700,000 and
800,000 VAX and AlphaServer machines over the past two decades, and HP has inherited an installed base
of customers that probably numbers in the hundreds of thousands. These companies may not buy the latest,
hottest technology every year, but they do buy upgrades and pay maintenance, and HP cannot turn that
down. HP is forcing HP 3000 customer to either migrate or go away--albeit with a very long time before
these HP minis are sunsetted--because its installed base is just not in the same league as the AlphaServer
and VAX base. It's a numbers game.
IBM has similarly sold approximately 250,000 System/3X minis and around 800,000 AS/400 and iSeries
midrange machines over the same time period, and has an installed customer base of around 250,000
unique customers these days. Let's do a little math here, just for fun. My guess is that, if you went to the
four different general managers of the Server Group and asked them how many unique customers they had,
they would tell you something like this:
xSeries: anywhere from 200,000 to 400,000
iSeries: 250,000
pSeries: 40,000
zSeries: 20,000
So it would look like IBM had a total of anywhere from 510,000 to 710,000 unique customers, right?
Wrong. These numbers would not account for the overlap where IBM was selling more than one server
type into an account. For instance, maybe half of the iSeries base is double-counted among the xSeries
customer base, too. (The other 30 percent of iSeries customers would, in this estimate, buy their Intel-based
servers from other vendors like HP, Compaq, and Dell.)
When you take all the double, triple, and quadruple counting out of these estimated eServer customer bases,
you would probably end up with a number that is closer to 400,000 truly unique eServer customers. And
with 250,000 of them being iSeries customers, you just can't push them too far. There's strength in that
number, and each and every one of you should remember that.
The problem is that there is not a good way to bring that collective power of the OS/400 customer base to
bear in helping to promote the OS/400 platform. The top 20 percent of IBM's customers, ranked by
revenue, determine where IBM goes or doesn't go and what it does or doesn't do. What we really need is an
active OS/400 union that has real input into IBM and can collectively set its goals for the platform that they
ultimately pay for. The iSeries Nation, sponsored by IBM
and the COMMON user group, as well as the many
individual OS/400 user groups, all have some input, to be sure. But it is not real power. If there were an
OS/400 union that engaged in what amounted to collective bargaining on behalf of those 250,000
customers, to try to change the iSeries for the better, would you participate? Who would run it? How would
it work? Mull this idea over, and let me know what you think, at tpm@itjungle.com.
|