by Timothy Prickett Morgan

In a shocking but not unexpected move, Hewlett-Packard last week killed off the HP 3000 proprietary midrange server line. The HP 3000 server line, which was launched in 1972 and is now known as the e3000, is almost as old as the IBM System/3X-AS/400-iSeries family of midrange servers and has an equally staunch and loyal following, particularly among manufacturers who use the machines for bookkeeping as well as for shop floor controllers. While the death of the HP 3000 line and its proprietary MPE/ix operating system should be cause for concern for OpenVMS customers over at Compaq, which is set to merge with HP early next year unless a proxy battle over the fate of the merger erupts, HP's sunsetting of the HP 3000 platform does not necessarily portend a similar fate for the venerable OS/400 platform. While the three big proprietary minicomputer lines from IBM, HP, and Digital (eaten by Compaq in 1998) do have much in common, they have very different customer bases and, more importantly, different prospects for the future.

The HP 3000 line was what got HP into the computer business back in the early 1970s. The machine and its MPE operating system are highly regarded as being stable and reliable. The strength of the HP 3000 business, where the company built up an installed base of about 60,000 machines by the late 1980s, is what gave HP the conviction and the means to create its own RISC processors and its own variant of the Unix operating system to compete against the then-fledgling company called Stanford University Network Microsystems, now known merely as Sun Microsystems. In the early 1990s, to help reduce its hardware development costs, HP moved its proprietary MPE operating system to the PA-RISC servers it created to run HP-UX, its Unix variant. Both Digital and IBM have since followed suit and done the same thing with their own proprietary operating systems. Although their RISC-based servers were designed to support proprietary and Unix environments (albeit not concurrently) from the get-go. And, in truth, IBM has ported AIX to the 64-bit PowerPC AS/400 platform, more than the other way around. As I have said before, the AS/400 Division, in creating the only working 64-bit PowerPC processors, saved IBM's cookies and has allowed Big Blue to come back in the Unix server market in a way that would not have been possible with the 32-bit processor designs that IBM's Austin RS/6000 labs put out. (Depending on Motorola to develop the 64-bit PowerPC 630 processor was, in retrospect, not a smart move.)

Having ported MPE from the proprietary HP 3000 processors to run on its 32-bit, and then 64-bit, PA-RISC chips, HP could continue to sell machines to existing customers who were loathe to port their applications to Unix or Windows servers. (Sound familiar?) While I am certain that keeping MPE alive on the PA-RISC platform does not cost HP much these days, HP is in a bind financially, and it is sunsetting the HP 3000 line in an effort to boost revenues of its Unix and Windows server lines while reducing costs. If it kills the HP 3000, that's one less operating system to have to support and enhance. If it kills the HP 3000, that's potentially a few billion dollars in Unix and Windows server sales over the next five years, when support for MPE will be dropped.

The real problem is that HP is in the midst of moving from the PA-RISC chip over to Intel's 64-bit Itanium processors. Back in 1996, when HP signed on as a design partner for the Itanium chips, it had no intention of selling servers using its own PA-RISC processors beyond 1999 or 2000. The current high-end Superdome and midrange N-class, entry A-class, and L-class HP 9000 Unix servers were designed to support the Itanium processors. But Intel bollixed up the design, the chip was late and underperformed, and that has forced HP to extend the life of the PA-RISC line. Intel seems to be working the kinks out with the McKinley and Madison generations of the Itanium processor, and HP is confident that it will soon be using these chips to support Unix, Windows, and Linux platforms on a unified server platform.

Hardware development is not the issue that forced HP to kill off the HP 3000 line. Software developed by HP, as well as third-party systems programs and applications is what this is all about. HP has spent the last three years porting HP-UX to Itanium, and has laid off the 150 engineers who did most of that work and, significantly, who could have made possible a port of MPE to the Itanium chip. It is hard to say what it would cost HP to move MPE to the Itanium processors, but it seems clear from HP's decision to kill that port, which was already underway, that it would not be a profitable move. Some HP developers who work on MPE have suggested that the company make the operating system an open source program so users and key HP developers can continue to support and enhance it on HP or other iron, but this seems unlikely. HP wants to steer HP 3000 customers, and steer them hard, to HP-UX, Windows, or Linux solutions running on its future Itanium server line.

Most significantly, sources at HP say that the ecosystem surrounding the HP 3000 and its MPE environment has degraded severely. Whereas IBM has sold close to 1 million System/3x-AS/400-iSeries boxes and has some 250,000 unique customers today, the HP 3000 base was probably never much larger than 100,000 to 200,000 machines and probably only had several tens of thousands of unique customers. Loyal HP 3000 users blame HP for killing off the 3000 sales force, for insufficient marketing, and for not keeping the platform up with the times. Because of neglect, the HP 3000 installed base has been declining in the past few years. As it now stands, HP will only sell new HP 3000 processors using PA-8600 and PA-8700 processors through November 2003. Customers with HP 3000 servers with modern 64-bit PA-RISC processors will be given the tools to install HP-UX on their machines, which HP is making available for free.

So what does this have to do with the OS/400 platform? Plenty. But the prognosis for OS/400 is a lot better than it ever was for MPE and a lot better than it will be for OpenVMS once HP and Compaq merge. For one thing, OS/400 and related database and systems programs bring in for IBM several hundred million dollars a year in revenue--my best guess is somewhere between $300 million and $400 million, depending on the year and the upgrade cycle. This software is extremely profitable, with somewhere around 85 percent gross margins.

Moreover, HP did not give MPE an HP-UX runtime environment or native Linux support within partitions. IBM has added the PASE AIX runtime and Linux in partitions to OS/400 servers, which means that AIX and Linux applications that would have never been ported to OS/400 can now run natively on the AS/400 and iSeries platforms. This technology difference--which is by no means above the technical capabilities of HP, which has several different kinds of logical partitions available on the HP 9000 platform- -has been crucial to the long-term economic survival of the OS/400 platform. Now, instead of having to port Unix or Linux applications natively to OS/400 using its compilers, IBM can simply take a program written for AIX or Linux and drop it into one of those environments on an iSeries server. Moreover, customers can run application servers for popular programs on the Integrated xSeries Server, which supports Windows NT and Windows 2000. OS/400 shops will continue to have options (which arguably are less attractive than actual native OS/400 support) because of IBM's inclusive yet controlling attitude to non-OS/400 operating systems on the AS/400 and iSeries platform. The real crime is that HP didn't see that it could have taken care of those MPE customers if it thought a little more outside of the box. Thankfully for OS/400 shops, the OS/400 community both inside and outside of IBM's Rochester iSeries development labs and its Somers marketing organization have been able to keep Big Blue moving ahead without giving up on the platform.

That said, the true measure of IBM's loyalty to the OS/400 platform will be annual sales and profits, just as it was for HP. The NUMA-Q line has been quietly killed off inside IBM, for instance, and no one even knows or cares except for a few big customers like Boeing that rely on those machines. So long as IBM can sell $3 billion or so in iSeries servers a year and a few hundred million dollars in OS/400 operating systems, the iSeries will be here to stay. There's probably quite a bit of wiggle room in those numbers, too. My best guess is that OS/400 only takes about $50 to $60 million a year to develop, and that IBM could sell only $2 billion a year in iSeries servers and give OS/400 away and still have a moderately profitable business. Only Buell Duncan, general manager of the MidMarket Server division, and his bosses know exactly where the break-even point is with the iSeries line, however, and you can bet that they will try to stay well above that point.

SPONSORED BY TANGO/04 NEW! Centralized iSeries and Windows application monitoring! VISUAL Message Center's graphical console monitors your entire iSeries and Windows server network for application error messages. * GET ALERTED instantly to all critical application errors in iSeries or Windows servers, by email, pager. . . before your business is affected * AUTOMATE iSeries message replies or command actions * SUPPORT iSeries users with screen capture, keyboard control * Graphically DEBUG iSeries production jobs, interactive or batch, even with remote source code--avoiding the need to recreate problems in testing * DIAGNOSE errors - see and edit LDA, QTEMP, LIBL on other jobs * Prompt and run commands on any running iSeries job to SOLVE LIVE PROBLEMS - preventing application downtime Call us at 800-545-9485 to request a free demo, or rush to our website (download available)! VISUAL Message Center from Tango/04 & SoftLanding Systems http://www.tango04.com/vmc ; info@tango04.com http://www.softlanding.com/vmc/400 ; info@softlanding.com

by Shannon O'Donnell

Inventive Designers, the Belgian software developer best known for its OfficeVision/400 replacement product, DTM/400, has announced a new release of EverGreen/400. EverGreen/400 is an industry-leading 5250 email product for the iSeries. Expanding on the already powerful array of email functions, this new version of EverGreen/400 comes with usability enhancements that reflect support for rapidly changing Internet standards, thus maintaining competitive, state-of-the-art technologies for iSeries shops worldwide.

EverGreen/400 V2R1 supports the Lightweight Directory Access Protocol (LDAP) for 5250 iSeries users and their administrators. LDAP is an Internet standard that allows a connection to any standard Internet or intranet address book on another server. This provides immediate access to the email addresses of customers, colleagues, and vendors.

Additional address book enhancements include the following:

* Multiple personal address books per user

* Import/export addresses to and from DB2/400 files

* Import/export addresses to and from CSV files

* Export address to vCard (.vcf)

For those shops using AnyMail/400 as their Internet email server, this new version of EverGreen/400 now supports automated replies and forwarding and has improved security. In fact, users can now configure their email to automatically respond to incoming messages with an "out of office" message, or they can have those messages forwarded to another email address.

Other enhancements include the following:

* Expanded subfolders--to enable detailed organization of inbox, sent mail, and archives

* Message templates--to design your own company forms, such as invoices and purchase orders

* Default directory for attachments--to limit or enhance user access to specific files only

* Display of HTML text and attachments