[Info-vax] OT: Arun Kishan

[John Wallace]

On Jan 11, 9:23 am, Michael Kraemer <M.Krae... at gsi.de> wrote:
> JF Mezei schrieb:
>
>
>
> > Lets not forget that for a company such as HP, it isn't just the chip
> > that is different, but also the systems.
>
> I don't get it why people believe in x86 as being the saviour.
> I can't remember any ailing OS ever having risen from
> the ashes by switching to x86 (or any other hardware platform).
> An OS has to survive by its own virtue.
> BeOS for example was cool when it ran on PPC boxen.
> Switched to x86 and went under.
> Data General's DG/UX ran on 88k, switched to x86 and went under.
> Etc, etc.
>
> It's not just that some code has to run natively on x86,
> the hardware built around it is at least as important.
> Look at OS/2 for example, it's x86 right from the start,
> but listening to the respective NGs it's pretty hard to find
> current x86 hardware to run it, in particular the coolest one,
> i.e. net- and notebooks. The same is true for
> Linux until very recently, when it became commercially
> interesting to offer netbooks with Linux rather than Windoze,
> i.e. to avoid the M$ tax on top of hardware costing $1xx.
> So even if (!) HP would bother to port VMS to x86,
> it'll most likely run on proprietary HP x86 boxen only,
> and in this case you're not much better off than with
> proprietary Itanics

VMS isn't about "cool", it's about business benefits. BeOS and DG/UX
were already struggling; switching to x86 did not cause them to fail,
but it didn't do them any long term good either.

There is little architectural difference between the innards of a
decent modern AMD64 server and a low end IA64 box [1]. At today's high
end of IA64, there are no fundamental architectural reasons why the
AMD64 boxes could not one day extend upwards into massive-memory
massive-SMP territory, but the AMD64 business model says that very
very few workloads need that kind of system, so why bother with the
development costs (especially if some other empire in the same company
is already catering for that market).

VMS survives today because people are interested in the business
benefits the VMS software brings. Subject to the massive-memory
massive-SMP caveat above, IA64 hardware for VMS brings no benefits,
only disadvantages, because if any of the players involved in a
potential VMS project is unconvinced about IA64 - the end user, the IT
department, the application provider, the consultants, etc - then
often the benefits of VMS are ignored and the project likely goes for
mainstream hardware (AMD64) with a mainstream OS (*ix, Windows).

If VMS was available on some subset of mainstream AMD64 hardware, that
problem vanishes. HP already make a very reputable set of AMD64
servers. They, or their Dell equivalents, are running enterprise-class
software in businesses around the world, and their compatible and
affordable little brothers are running smaller businesses too.

You wouldn't need VMS support on all of HP's AMD64 boxes, just a
representative few. And if as a bonus VMS works but is unsupported on
other AMD64 family members, then that's the "cheap machines for VMS
development" issue sorted.

I'm not holding my breath though, especially as it would leave the HP-
UX community (alone) picking up the whole cost of IA64 development,
and that might not go down well.

[1] If there are, I'm sure readers will be happy to be enlightened.