HP hardware: bargain, but needed now?
Where Everybody Knows Your CPUNAME

How Ending Support Might Change Things

If the above subject seems obvious, then the story of the HP 3000 and MPE has had moments to refute it, as well as prove it. Hewlett-Packard considered the end of its vendor-priced support to be the ultimate change in 3000 ownership. If HP wouldn't support MPE and the 3000, who'd use it?

That one is filed as a refute -- several thousand companies have relied on 3000s and MPE over the four-plus years since full HP support ended. Even as a government-required archival system, the computer outlasted the end of HP support.

CliffBut in proof of ending support as a trigger of change, we offer the case of the disappearing database. No, not IMAGE, still wired to this day with elegance into the MPE filesystem and 3000s. No, we're examining Oracle here. Many IT managers consider Oracle to be the industry leader. So if its support drove off the 3000 cliff, and so dropped off for MPE after Y2K, didn't that deal a crashing blow to the user community?

One manager who wants to remain anonymous, but still tends to a 3000, told us this week that he believes it was true. "I asked HP people at a trade show if they had heard how Oracle, recently in court and in the news, began the demise of MPE -- when in a previous pre-Sun business decision, they announced end of support for Oracle on MPE?"

Yes, the end of this support did change the 3000's future -- at HP. In the early 1990s HP was hoping that IMAGE would become only one of several database options for the servers, and so it tried to unbundle the custom-tailored IMAGE from MPE. This was meant to make room for the likes of high-dollar Oracle, or other databases which had not made the port to the 3000. HP wished they would do so. Hewlett-Packard's 3000 group pined for SQL Server on MPE.

But Oracle never was thrilled to be part of the 3000 ecosystem. There was so much more profit to be had in the Unix world, or up on IBM mainframes. In 1985 I was reporting on rumors that Oracle was moving to what we were still calling MPE V at the time. The Oracle VP I reached had a question for me. "Why in the world would we do that?"

About four years of market time changed this approach, but in '85 Oracle wondered why in the world any vendor would offer an MPE-ready database while IMAGE was included with every 3000. How could Oracle compete with that value? It wasn't like HP was reducing the prices for 3000s which shipped with no database. Enter, after Oracle's arrival on the 3000, the scheme to unbundle IMAGE -- and the customers' revolt at a public roundtable over this strategy.

Oracle trod its path away from MPE after many versions of the database were built and rejiggered to try to match the IMAGE advantage: speed, and the compatibility with the catalog of HP 3000 applications and tools. While that latter group of software bled away -- cut back by the vendor's forecast of an ailing 3000 ecosystem -- Oracle updated Oracle/iX less and less frequently.

That might be the same situation HP's other enterprise environment will experience with the loss of Oracle support. Oracle didn't want to continue with the successor to the 3000's hardware architecture, Itanium. The courts forced this otherwise, but the last period of that game is yet to be played. 

Will the end of Oracle support for HP-UX change things? Of course. Will it be a fatal blow? More fatal than the Oracle evaporation from MPE. Few applications ever relied on Oracle on 3000s. It didn't force a transition. Oracle never got traction as a database for in-house apps. That is a different situation than the reality for HP's Unix servers. But one third-party software platform like Oracle never sinks an enterprise ship on its own. It takes a vendor to deliver the coup de gras -- and even then, the demise can be years away.

However, for a company migrating to a new application, the existence of Oracle on that platform can be welcome news. Oracle administrators and developers are in rich supply. Replacing one who quits to become a coffee cart owner or gun-shop salesman is easy. Oracle is everywhere, and thousands of applications rely on it. In a 3000 migration situation, the absence of Oracle support is more reason to change an application from something in-house to something else. "We always wanted to walk away from those old apps," a MANMAN manager said in a user group. "HP ending support for the 3000 gave us the opening."

The ruler to measure change by: the critical nature of the element whose support is ending. Oracle isn't MPE, but it might as well be for any company that considers itself an Oracle customer first, with a few HP 3000s as outliers. As we can see today, support for IMAGE has not ended, thanks to independent companies. Oracle on HP-UX support -- like the 3000's, without new development -- might trigger changes.

How HP reacts to the eventual Oracle departure this time around, compared to its reaction in the 1990s, will determine how much support can change things. The greatest change from ending support takes place in vendor-built environments which have receding application ecosystems. What's the trend for application vendor enhancements in your target environment?