At yesterday's demonstration of the current HPA/3000 virtualization engine -- you'd know it as The Emulator -- tech success was in abundance. Everything that the product manager Paul Taffel showed off during the spring CAMUS User Group meeting worked as expected. One hour of demo without a single crash. Glance, third party database tools, even something as esoteric as the OCTCOMP object code translator that HP built, the one that ensures that Classic 3000 programs will run on the PA-RISC systems of the modern era.
That last item was no slower than it behaves under HP's 3000 iron. Taffel was showing off the N-Class version of HPA/3000, and he was doing the demo on a $1,300 PC using a paltry 4GB of RAM within 16GB on a Linux PC.
My test Linux system has 16 GB of memory (although we only recommend or need 8 GB for the A400 emulator). I was running our N4000-100-750 emulator, with 2 GB of memory available to the virtual HP 3000, and I mentioned that I was actually only using about 4 GB of memory at the time (including Linux overhead). I'm not sure that increasing the memory allocated to the virtual HP 3000 would have resulted in any noticeable speed-up, at least during the relatively low load tests that I was performing.
The marvel of the HPA/3000 design is that it has no measurable ceiling for top performance. Intel will keep improving its chip speeds. That puts more horsepower at the command of this engine.
Licensing advocacy might speed up sales. Some vendors are going to want to test for the market's only 3000 emulator and need to recover the lab costs. Others see a need for more real-world tests. HPA/3000 isn't a software-software interaction, however. Taffel says there's no MPE/iX emulation going on in HPA/3000. Every feature of the 3000's OS operates the same, right down to intrinsics. Yes, even the end-of-2027 date bug exists in the emulated solution.
Terry Floyd suggested that an organization like the storied SIGSOFTVEND could assist in getting 3000 apps and essential tools certified. Taffel called the business that vendors could protect via emulator customers "like money for nothing. It's giving these vendors another few years of software support business."
That's true, unless few members of a vendor's supported customer base have mentioned the emulator. These vendors need to be convinced, by some advocate, that it's good business to include HPA/3000 sites on their approved list. Without any evidence they're going to lose money if they don't do HPA/3000 tests, vendors could play the short game and aim tech resources elsewhere. A 3000 software vendor organization is a good idea anyway, but the HPA/3000 gives it a real business focus. SIGSOFTVEND did its work with no overhead to speak of -- and that group was tracing the impact of HP's changes to MPE/iX. VEsoft's founder knows embracing HPA/3000 is simpler.
Supported customers do leave a vendor, after awhile. But hesitation over emulator certification may be a sign that a vendor is looking at other places to invest right now, at least until the HPA/3000 gains some customer traction. This reticence represents the cost which HP levied by stalling the Stromasys product for five years. A green light when Stromasys was ready in 2002 would have yielded a product by 2005, or even 2008. At either time, software suppliers would've had a lot more customer support contracts to protect with emulator certifications. Taffel called that delay a tragedy, but he's on the job to create happier endings.
"It's a tragedy if you care about extending the HP 3000 lifetime," he said. "But I'm not singling anyone out for blame. That's just the way things worked out." Stromasys CTO Robert Boers was more explicit last fall about the source of the delay. Everyone agrees that selling this product four years ago would have been easier. As it turned out, HP had lost more than half its migrating sites to other vendors by the time it started to work with Stromasys on HPA/3000.
A test suite for software products need not be extensive, in order to keep them from being expensive. This product does nothing more than make Intel processors behave like PA-RISC chips. VEsoft's Vladimir Volokh said using MPEX on this platform -- a product which literally extends MPE -- didn't require software-software level testing, as far as he was concerned.
If an HPA/3000 prospect has got all in-house code -- and only needs the surround tools which are already certified by third parties -- then they're more likely to arrive on the emulator. Licensing, or even certification, can be very important to closing a sale. Some products like MANMAN don't even look for a HPSUSAN number. Floyd said that in the case of that app suite, it comes down to a matter of ethics. We would add, "and what your auditor expects about licensing."
Stromasys has never launched a product into a market where the vendor was totally absent, so long-gone that you may wonder if there's an HP 3000 license transfer mechanism running anymore. (HP says yes, and we've had reports from resellers who use it.) The customers think it exists, mostly. That's why the first-wave companies are going to be so important to the launch of HPA/3000. People want to see how others are handling the business matters of using an emulator -- and those decisions on licensing may be the crossroads of whether anyone will spend $25K to $100K on HPA/3000.
But at least after yesterday's meeting the world knows it works with MPEX and DBGENERAL and SHOWCLKS. It looks great, and there was a magic moment where an HP 3000 boot volume was duplicated, or compressed, using Ubuntu Linux. Taffel says at some point the product will be called a 1.0, instead of the prerelease 0.8.
Advocacy from someone in the community with business leadership could help include crucial tools on par with MB Foster's connectivity software, or even PowerHouse and Speedware. VEsoft's in a vanguard here with several other software companies fanned out with a few thousand sites. Stromasys can't do certification advocacy as well as a software supplier with deep roots in the 3000 community. There's selling of HPA/3000 to be done on more than one level. Without advocacy, it may be every HPA/3000 customer for themselves in arranging to use the solution with their third party software.