HT-2021-000363 - [2025] EWHC 532 (TCC)

In 1998, Jan Jaeger, now the Chief Technology Officer and Chief Software Architect at LzLabs, with his colleague, Roger Bowler, developed “Hercules”. Hercules was a mainframe emulator, which supported, on an x86 CPU, functions equivalent to those provided by IBM mainframe hardware. The purpose of Hercules was to enable programs designed to run on the mainframe to be executed on a personal computer with an x86 processor by executing z/Architecture instructions, and implementing that instruction set on an x86 machine. Although a significant achievement, Hercules could only emulate the mainframe hardware; it did not replace the mainframe runtime environment. Consequentially, it could not be used to run a mainframe application directly; it was necessary to install on Hercules an operating system, such as Linux for zSeries, to enable it to run such an application.

The development of Hercules by Mr Bowler and Mr Jaeger was achieved by implementing x86 versions of each of the machine instructions executed by mainframe CPUs. This task was facilitated, in part, using documented instructions, and in part, using reverse engineering for undocumented instructions, as explained by Mr Jaeger in cross-examination:

“Q. Mr Jaeger, a fair description of the work you carried out was reverse-engineering on the mainframe, wasn't it?

A. I -- well, you can call it reverse-engineering, fine. I don't know what you mean exactly by that when it was most -- we were not interested in how it worked, it was just – it’s just like what we’ve done with the SDM. We want to know interfaces. A single instruction is also an interface. We just needed to know the interface.

Q. Mr Jaeger, we're concerned here with undocumented instructions on the mainframe processor, aren't we?

A. Yes.

Q. Are you seriously saying that you consider those to be interfaces?

A. They're interfaces into the hardware, absolutely.

Q. I see. Which mainframe did you use to do that work?

A. We had a mainframe at the ING Bank. It was running 14 z/OS, z/VM, Linux, it was running everything.

Q. I see. And that was the mainframe that was used?

A. Yeah.”

Mr Jaeger’s evidence was that some of the Hercules code was incorporated into the product that was developed into the SDM. He agreed that the SDM needed that compatibility layer with the processor instruction set to run the instructions on an x86 processor but denied that any reverse engineered instructions from Hercules were used in the SDM.