[Info-vax] VMS port to x86

John Wallace johnwallace4 at yahoo.co.uk
Mon Mar 26 17:15:20 EDT 2012


On Mar 26, 9:40 pm, JF Mezei <jfmezei.spam... at vaxination.ca> wrote:
> Johnny Billquist wrote:
> > Yes, and that is not properly "bare metal". Having Linux on which your
> > emulator runs, means it's running under Linux, not on the bare metal.
> > How hard can it be?
>
> Except for IBM's LPARS (or whatever it is called) and Galaxy/Wildfire
> class Alphas, are there hypervisors that allow multiple instances of an
> OS directly on the hardware without some sort of host operating system ?
>
> In the case of HP, the hypervisor they use on IA64 runs an instance of
> HP-UX which then hosts other instances as glorified applications.
>
> Anyone know what VMware really is ? It wouldn't susprise me if it had
> linux under the hood.
>
> Windows' offering is of course based on windows.

Yes various directly-bootable flavours of VMware have (historically)
had Linux under the hood. If I remember rightly they have ended up in
court because they weren't sticking to the licence (sorry, can't find
details right now).

I think as a result of those legal cases, VMware now runs on top of
something allegedly home grown which allegedly isn't Linux (so no GPL
to worry about) but from the bits I've seen written about, to anyone
with a bit of a clue it looks remarkably similar to what a kernel code
writer would see as Linux.

The difference between genuine "bare metal" and "thin OS" may matter
(not to everyone, but to some people). To some people it may matter
from a legal point of view, as VMware found out. And to some it may
matter from a practical point of view. E.g. to customers who are
retaining VMS because of its history of stability, when you put it on
top of a layer that introduces its own bugs and vulnerabilities. OK a
thinner layer than Windows, with fewer bugs than Windows, but not an
invisible layer.

The "not really bare metal" approach gives customers the flexibility
to install their VMS-machine emulator on a commodity box on which the
"not really bare metal" OS will almost certainly run OK much of the
time, but the "not really bare metal" OS will almost certainly not
have been tested and qualified to the level people used to expect from
VMS.

This "not really bare metal" OS will (presumably, as it's a thin
layer) not have the error recovery and diagnosis features which VMS
users have come to expect, right? If an intermittent disk error
occurs, how is it diagnosed? VMS managers running VMS on real hardware
know how. What happens in this environment, is there an error log in
the "not really bare metal" OS? Or does that kind of diagnosis in this
kind of environment require telepathetic skills?

There are lots of cases running simple legacy VMS apps where this kind
of stuff won't matter. Fine. But VMS often lives on in more complex
cases too, where it might matter more than is being admitted...



More information about the Info-vax mailing list