[Info-vax] Some questions on software for VMS 7.3 VAX

[Stephen Hoffman]

On 2016-01-13 00:39:12 +0000, David Froble said:

> abrsvc wrote:
>> Most of the VAX clients I deal with still (there are a few), remain 
>> there due mostly to hardware constraints.  There are many sites that 
>> have custom hardware connected to the VAX that has no direct 
>> replacement on either Alpha or IA64.  Since the code (and machine) 
>> continue to work without failures, it continues.  While I am concerned 
>> that hardware of this age is likely to fail at any time, these clients 
>> continue to depend on it.
>> In these cases, "new" stuff really has no value...
> 
> Well when (if) VMS gets to x86, perhaps some of the custom stuff could 
> be easily replaced, and such customers would then have a new life for 
> their applications.
> 
> Not knowing what the custom devices are, can't determine the difficulty 
> in replacements.

That's not likely.   The I/O buses aren't going to change all that much 
from what's available now with Itanium: mostly PCIe and USB, or 
outboard PLCs or embedded boards.

Migrations are probably going to FPGA on PCIe, or to some PLC or 
analog, or wholesale replacement of the OpenVMS end with whatever is 
current when whatever is out on the far end of the connection — what's 
usually the expensive or hard-to-change part of the configuration — 
gets replaced.

What freezes more than a few of these cases involves both the weird 
network protocols — and which tend to be completely bizarre, and more 
than a little timing sensitive — and recreating the device driver 
interfaces that are more of a problem, if you can't disrupt the apps on 
the OpenVMS end.    That, and what they have is working.    There's not 
much point in upgrading or replacing OpenVMS for these folks, in other 
words.   If they could even do it.

PLC communications protocols commonly used back in the 1980s were... demented.

For more than a few folks, the factory or the equipment test frame or 
the crane or the turbine or whatever is out on the other end of the 
connection will eventually get replaced.   Until then, the VMS box is 
little more than a peripheral of that other function.

These sites are not going to be VSI customers, either.   Not until and 
unless the third-party providers and integrators that are inevitably 
involved here decide to use OpenVMS for their embedded computing, too.

This is kind of like replacing the vehicle in-dash entertainment 
systems that are being sold in more than a few cars — possible, but a 
whole lot of work for not much gain, and will usually only happen when 
(if) you buy another car.   The cases that Dan and I and others have 
seen or have worked on aren't that far off what the IoT folks will be 
seeing — and the rest of us — all over the place, either.


-- 
Pure Personal Opinion | HoffmanLabs LLC