[Info-vax] VMS port to x86

[John Wallace]

On May 31, 5:42 pm, Keith Parris <keithparris_deletet... at yahoo.com>
wrote:
> On 5/31/2012 4:51 AM, JKB wrote:
>
> >    If you want to keep VMS alive, you must have a VMS port to amd64.
>
> I agree. x86-64. And ARM might not be a bad 2nd target.
>
> >    I'm not sure that HP would start this kind of development.
>
> We know fromhttp://www.oracle.com/us/corporate/features/7-1623010.pdf
> that as of the end of 2008 Martin Fink said the VMS business was less
> than $100M (although one could argue he probably wasn't including
> Services, Storage, Networking, etc.). We know fromhttp://www.oracle.com/us/corporate/features/10-1623014.pdfthat HP-UX
> was estimated to cost more than $100M to port beyond first boot on x86.
> Perhaps because of the work already done during the Itanium port to make
> OpenVMS more portable and because it's already Little-Endian, OpenVMS
> might be significantly less expensive than HP-UX to port. Still, can the
> OpenVMS organization realistically afford to spend some number that's
> probably on the order of at least one year's total revenues to do an x86
> port? There's also fear that in another architecture migration, even
> more ISVs (and thus customers) would be lost. If an x86 port did happen,
> I'm guessing it would have to be based on HP's goodwill and commitment
> to its OpenVMS customers, not because of its financial attractiveness.
>
> >       Thus, conclusion is very simple, we have to start to write a VMS clone
>
> VMS is viewed by the marketplace in general as legacy, out of date,
> old-fashioned, or at least irrelevant. Even if the existing code is
> open-sourced, or HP ports OpenVMS to x86, it will still have the same
> perception in the marketplace. (We here know that OpenVMS has valuable
> attributes such as 4-ring security and high availability with shadowing
> and clustering and scalability with clustering that are ideal for
> current marketplace needs, but most everyone else can't see past the
> "Out-Of-Fashion" label that's been stamped on it. So it's been harder
> lately to pick up new customers and new ISVs.)
>
> To be fair, most proprietary UNIX flavors are also seen in much the same
> light. Linux, on the other hand (although we all know it is just UNIX,
> warmed over) is seen as new and exciting.
>
> An open source VMS clone would likewise be seen as something new and
> exciting and interesting, although we here know it would most likely
> simply deliver the same types of value and capabilities provided by the
> existing proprietary OpenVMS product in an open source (thus free in
> purchase price) and open (in terms of vendor independence and
> development freedom) distribution model.
>
> An open source VMS clone would fit well with HP's direction with Project
> Odyssey, which aims to provide as much as possible of the value of
> HP-UX, NonStop, and OpenVMS on x86 server hardware.

"fear that in another architecture migration, even more ISVs (and thus
customers) would be lost."

Surely one could make a strong case in the opposite direction too,
given the circumstances? An AMD64 port could be a positive for VMS,
actually making it possible to attract previously unaddressable
customers, and giving the existing IA64 base a feeling there is
somewhere comfortable, somewhere widely supported, reasonably
affordable, and with some kind of future, for them to go once IA64
inevitably goes EOL, and also somewhere where low end systems are
actually available at sensible prices for developers, testing, etc. I
do agree that migrating to (say) MIPS64 or SPARC or ARM or maybe even
Power would be madness. ARM might be interesting and fun madness, but
madness all the same.

If a migration to AMD64 were to be announced, ISVs and customers
(existing and potential) would once again see signs of genuine
commitment to and investment in VMS, something which has been absent
in recent years. ISVs and customers would finally see promise of VMS
running on (presumably a qualified subset of?) the server family that
the most of the world already relies on, Compaq Proliant (still
selling substantially more than the Dell competition, both in units
and in value). What could possibly go wrong?

Actually one of the biggest things that could perhaps go wrong in the
foreseeable future might be having HP HQ (rather than someone with a
real positive reputation in enterprise computing) being responsible.
I'm not sure HP HQ's recent track record with stuff other than
printers would be viewed as positive or negative by existing or
potential VMS ISVs and customers.

If not HP, who else's name might be in the frame? Are there any other
companies still out there that can still demonstrably run a business
capable of supporting more than one OS on more than one chip family?