Reader Feedback on New Systems and QuickTransit Emulator

October 3, 2011 Timothy Prickett Morgan

Timothy:

Thumbs up on the lead-off article Start Planning For New Systems Now. Thorough but useful at every turn, practical rather than a tech spew of speeds and feeds, and a great bit of tactical advice regarding negotiation on a Power7 in light of the upcoming Power7+.

Just wanted to let you know I enjoyed it and will be referencing it where relevant in some LinkedIn groups I participate in.

–MB

That’s why you keep me around. Glad to be of use.

–TPM

So did Transitive decide to quietly escape IBM’s wrath by being absorbed by them? It really seems odd that an emulator of the magnitude of QuickTransit could escape vendors’ potential litigation or the reverse, their support. Why spend 10 years on a product (probably on the taxpayers and alumni dime) and then cave in to IBM, who obviously wasn’t going to sell it?

That said, how does TurboHercules avoid the same fate?

That was always a bit of a puzzler to me as I learned about QuickTransit and saw how it was being implemented and sold. I said from the very beginning the plan was to have system vendors all license QuickTransit to raid each others’ system bases, and the brilliance of the scheme was to sell bullets to all sides in the systems wars. But maybe the bullets don’t work as advertised, or they require more expensive guns than we know of. All I can tell you is that Apple sure made a smooth transition with the emulator.

I think Big Blue figured out that its mainframe and Power Systems businesses had the most to lose and it was cheaper to buy Transitive than to allow it to fall into the enemy hands of Oracle, Hewlett-Packard, or Dell. HP could probably use QuickTransit right about now to emulate PA-RISC and Itanium code on X86 servers, and Dell potentially could have the most to gain, with no system platform of its own to protect. I think a QuickTransit environment running inside of a virtual machine on X86 iron would be particularly powerful, and better still, IBM could start putting hooks for it into its iron and do a very good job emulating Sparc or Itanium applications on Power–or mainframe apps on Power, for that matter. And if you think this approach is crazy, then I suggest you take a gander at the “Godson” MIPS processor being created by the Chinese government, which is embedding QEMU hardware emulation functions inside the chip so it can emulate X86 code on MIPS processors. The early tests show about a 30 percent performance penalty running X86 apps in emulated mode on the Godson-3C processors, which will ship next year.

As for TurboHercules, the company founded to provide commercial support for the open source Hercules mainframe emulator, the situation is complex. In August, the French company dropped its complaint against IBM to the antitrust authorities in the European Union back and took its website offline “to explore new business opportunities.” Hercules allows for X86 and Itanium processors to emulate mainframe processors since the dawn of time and is itself written in C and distributed under the Q Public License. It is harder to kill Hercules because it is open source, but it is also very difficult to get permission from IBM to run MVS, OS/390, and z/OS licenses on top of Hercules. So you can’t kill Hercules, but it can’t exactly live, either. Maybe they should have called it Prometheus?

–TPM