[Info-vax] x86-64-based Superdomes on the way (says HP)

[Phillip Helbig---undress to reply]

In article <4ed1662c$0$2234$c3e8da3$76a7c58f at news.astraweb.com>, JF
Mezei <jfmezei.spamnot at vaxination.ca> writes: 

> At this point, with HP basically having confirmed they are throwing in
> the towel on BCS hoping only to slow the decline of that unit, my guess
> is that HP will expand its VMS poslicy of just trying to cater to the
> existing customer base to HP-UX and NSK.

Why would they try to expand it?  They haven't done that for decades.  
And why now, if as you believe they want to wind it down?

> However, when IA64 is discontinued, I think BCS will remain, mostly for
> NSK. Even if the functionality of NSK is ported to Linux on 8086, HP wll
> still need to build fault tolerant systems to run that version of Linux.

Maybe not.  The world has changed.  In many (though perhaps not all) 
applications, reliability, availability etc were very important.  Today, 
many people are happy to trade that for raw speed.

> On the other side of the coin, perhaps Whitman sent this trial balloon

Doubtful.

> out to see if the HP-UX/VMS/NSK customer base will revolt 

The customers have always said that they want VMS to continue.  What 
would a "revolt" consist of?

> and demand
> their OS be ported to 8086 servers 

Most people don't see the the logic "if it runs on 8086 [or whatever]
then it is much more likely not to die".  Yes, I can see some advantages
of VMS on commodity hardware (which was what Itanium was supposed to
be), and perhaps modern Intel chips are more viable in this respect than 
they were in the past, but even if that happens, the money to keep VMS 
running has to come from somewhere.

> or tell HP exactly what features HP
> needs to add to Linux to make it possible for customers to migrate from
> HP-UX , NSK and VMS to Linux.

A non-starter.  While the fact that linux is free (at least if you don't
value your time) might appeal to some poor people and to the
anarcho-hacker crowd, most big users of VMS (i.e. those who, now and in 
the future, would actually be paying for it), if they move to linux at 
all, don't do so to save costs, since they would still have an expensive 
maintenance contract with Red Hat or whomever.  Rather, they would move 
since they are then not locked in to a particular vendor.  "HP adding 
features to linux", i.e. some sort of HP linux, doesn't free them from 
this captivity.  It would have to go into the general linux 
distributions, then the question arises why HP should fund that.  Anyone 
who finds himself unable to continue because HP stopped support of his 
favourite platform will have as a very important goal not to be locked 
into one vendor in the future.

> Perhpas it is time for Hoff to port VMS clustering to Linux.

Clustering is important, and AFAIK no other system can do what VMS could 
do already 20 years ago, but it is one relatively small part of VMS.