Vendors preserving 3000s for historic use
HP's 3000 software practice once wide open

What becomes legacy: everything eventually

What can you do about it? Embrace the virtualized future for any platform you use, or migrate onto. It's the only way to keep the business of the old republic vital.

Wars-the-old-republic2When Stromasys showed off its first test runs of the HPA/3000 Charon emulator, the company pointed out that the market for virtualization is only getting larger. Some companies have retained environments no longer supported by the vendor. While HP comes to mind here, this is also true of Microsoft. Some people are surprised when they learn MPE/iX drives manufacturing companies as large as Measurement Specialties. That's only a few dozen servers, however. We've heard of one organization that has more than 3,000 Windows NT systems to support.

That story comes from Robert Boers, the CTO of Stromasys who makes a compelling case for virtualization of all environments, whether a dated Windows release or the classic MPE/iX. "The industry is looking away from a growing problem of legacy systems," he said, "one that's got nothing to do with HP or VAX and Alpha servers." Stromasys has sold more than 4,000 installations of those DEC systems, developing a business model and tech design they're applying to HPA/3000. He cited a military organization "who have 3,500 Windows NT 4 servers. They can't run that stuff anymore on a modern [hardware] platform."

While Boers said this legacy wave has little to do with the HP 3000 opportunity, the issues remain the same. Everything which IT purchases has an end-of-life date coming from the vendor. Once that day arrives the IT customer can choose to go independent. But at some point a PA-RISC system like the 3000 or old HP 9000s, or the Itanium servers hosting HP-UX, it all becomes a legacy system -- even the relatively nouveau platform of Windows NT. All of these computers host business critical applications. Boers said the average lifespan of such an app is now 22.5 years. It's not tough to find in-house software that was created more than 20 years ago in your community. What's become harder to do is find gear to keep it running which has room to grow in performance and connectivity.

"There's still more and more legacy systems around," Boers said.  "if you look at the growth of thge worldwide installed base of computers, there's a very interesting period between 1980 and 1990. The worldwide installed base multiplied by a factor of 22 times. The IT installed base kept growing, but it has never hit a factor of 22 in any single decade since then."

"If you combine the two numbers -- 20-30 years of application life, and a growth of 22 times -- over the next decade we will see an unusually large number of business applications which are clearly at the end of life. It's accentuated by the growth bubble of the 80s. It's the equivalent of the post-WW II birth rate." IT managers put the HP 3000 on the map as a popular enterprise destination during that decade. Unix wasn't an option and Windows didn't exist in a practical release.

Whether you homestead or migrate into the future, this baby boom of IT will impact your plans. "We're going into a decade where there will be an extreme need for replacing something to keep these applications running, or find a more efficient way of replacing them. There is not enough time and money in the world or not enough people to rewrite the applications or put them into SAP. This symbolizes there have to be other things to be done with legacy applications."

"We simply can't cope with them over the next 10 years. And very few people know that, because the average of the computer salesman is less than the time when these [legacy] machines were created. They don't know about them, they don't like them -- and they have no clue how important these machines are."

Boers waxes philosophical about the situation once he starts talking about caring for legacy technology. "In the last 200 years or so, every time something was really needed in society to keep it moving, it was invented," he said. "Sometimes multiple people invented multiple things in different places in the world." At this point I think about the wave of minicomputers built in the 1980s from Wang, DEC, Data General, NCR, Burroughs, Sun, HP, IBM and so many more. So many have seen the end of their vendor life. Some of those are supported in virtualization, even today.

Boers said he believes that if the IT industry is maturing -- "which I hope it does, it will come up with newer technologies to deal with this growing problem of legacy software. I'll be less modest: one of the things we at Stromasys have invented is a little bit of that newer technology." The IT industry needs to spend more time helping customers with such dilemmas, he added.

The legacy market will be getting larger once Oracle sets down the end-of-life for its database on the Itanium servers, he said. Oracle's engineering for the RDB database has been done for years on the Stromasys emulators, he said. Over in that DEC community, Oracle provides free transfer licenses for this database onto the Stromasys emulator. Oracle has a presentation "that shows that the RdB database runs about 10 times faster on our emulator than on the original Alpha hardware," Boers said.

In a spot of irony, those Alpha customers were being forced to move to Itanium servers to preserve their environments, although there was a lot of re-engineering to make that migration a success. Oracle's talk of "terminal releases" of Rdb reminds every IT planner that Father Time wins outruns every technology eventually. But Oracle's support of a hardware emulator for Rdb helps make a case for choosing virtualization as a starting point, rather than a migration target. VMware has thorough IT buy-in as an enterprise solution as well as a future on the HPA/3000 solution. Buying specialized servers might build an ecosystem for a vendor to develop. But one business decision like Oracle's -- built upon competition instead of technical ability -- can sweep an ecosystem into hibernation. Although HP's biggest competitor, IBM, has usually given customers a way to protect applications, HP is only learning this now. The prospect of being hibernated doesn't seem to keep HP's customers from buying without a thought for a multiple-decade app lifespan, though.

"Large users of IT technology should be more vocal," Boers said. "They seem to swallow everything the industry does. I am puzzled why large IT consumers of tons of HP systems do this. They should be technically savvy enough to tell HP, "It's nice to go to a new platform -- but are there other ways in which you can sustain my existing business critical applications? A company like IBM has done this forever. In between companies I worked for the biggest bank in the Netherlands. On the Series 360s they had emulation mode for the older applications. They never had a need to replace them."