One of the prime reasons for extending 3000 application life is investment protection. It is difficult to justify, however, if a company continues to grow while its hardware performance remains capped at 2003 levels That's the incredible hamstring that most MPE/iX applications labor under. Aside from refusing to put MPE/iX onto Itanium chips, there was a fresh generation of PA-RISC processors available to HP by the middle of the last decade. But those PA-8900s were never employed in 3000s, just 9000s. Then there's the matter of hardware down-clocking. It's a feature not included in 3000 emulation.
Owners of 3000s probably know their systems were hobbled by HP during the design of the ultimate generation of the servers. They should also know that protecting their application investments with an emulator eliminates that hurdle to modern-day performance. The recommended top end today is an Intel i7. Whatever comes next will be available to keep MPE application performance growing.
In the wake of the just-completed VMworld show, it's easy to see that virtualization -- the other name for emulation -- is mainstream technology by now. Five years ago this month, though, we interviewed the Stromasys CTO Robert Boers about the design goals for the Charon emulator for PA-RISC 3000s.
Is your emulation going to get rid of the slowdown code that has hamstrung PA-RISC processors on 3000s?
We’re not using that. They’ve clocked them down to the equivalent of 55 MHz on the low-end models. HP actually had a back-door to allow their support people to turn up the performance if they were in a hurry. We’re actually building an accelerator.
The reports of Charon use in the field do not include any notes on clipped performance. The company made good on its promises of acceleration -- an area that can be enhanced by more upgrading hosting hardware, too.
But there's that hardware future that looks brighter. An N-Class server might not be as costly an investment burden as it once was, but emulation offers more growth. We asked Boers (who's retired now after a mainstream management team adopted the Stromasys solution) whether the clock would run out for a meaningful product release of Charon.
Some in the community say too much time has passed to make this a relevant product. What’s your take?
To tell you the truth, HP’s been pretty slow. I feel concerned, because we should have been ready much earlier. We’ve been waiting about a year until we got an agreement on the Processor Dependent Code information, because HP's overriding worry was the ability to run HP-UX. What concerns me is that there’s only about a half year left to get additional HP licenses. We might have a working beta by then, but not much more by the end of the year.
The emulator started to generate references by 2012, about two years after HP halted its additional license sales. As it turns out, MPE/iX licenses are still out there for sale in the form of rock-bottom used HP 3000 hardware prices. HP recognizes this value of these licenses. That's why it's operating a license transfer business almost 12 years after it sold the last of those down-clocked 3000 servers.