[Info-vax] Comment on the future of OpenVMS

[JF Mezei]

Keith Parris wrote:

> VMS customers are loyal.
> 
> Gartner said:
> 1) "If there is already investment in rival Unix versions," and
> 2) "HP has 'inherited' an Alpha user through the Compaq acquisition," 
> and if, as a result of that, it happens that "there is limited loyalty 
> to the company"
> then "vendors such as IBM or Sun Microsystems will be better-positioned 
> to benefit."

The way I had remembered it was that Gartner said that inherited
customers had little/no loyalty to HP, but VMS customers acquired by HP
would. Problem is that HP didn't do much to acquire NEW VMS customers.


>> with a scaled down el-cheapo VMS maintenance team in India
> 
> This is an insult to the hundreds of fine VMS Engineering folks in 
> India, and is uncalled-for.

No, it is not an insult to them. It is an isult to HP. There is nothing
wrong with the people in India, and yes, they provide very good value
for the money.

But when you have a high quality OS with very experienced crew to
develop/maintain it, replacing almost all of it in one shot is not
smart. And like it or not, HP didn't make this change to improve the
product, they made the change to lower costs.

So the "el cheapo" applies because the driver for this move was to hire
cheap labour to do the work.  If they had to pay the new guys wages
equal to those paid in the USA, would they have moved VMS to india ?

This is no way reflects on intelligence or capabilities of the people
there.

But just as people criticise Microsoft for hiring newbies with little
experience in writing solid code, the same applies to HP hiring newbies
who have little experience with VMS.

Had this been more of a transition instead of of a move with much more
time to transfer knowledge/experience, then there would be no
accusations of Hp replacing VMS engineering with a bunch of newbies.

Yes, there is a small core group with experience. But when you make a
massive staffing change as what happened, that core group would be
overwhelmed with their need to train the new guys, and if they don't
train them properly, you will end up with inferior quality code as
happened with Microsoft.

This whether the newbies are Indians, New Yorkers, Alaskans or
Californians is the same: make a massive staff change all of a sudden,
and you lose a lot of the knowledge and expertise because there is no
time to transfer it all.



> There are still very valid reasons for people to buy VMS.

We were discussing Oracle (or other) buying VMS from HP. Not some
customer buying a VMS licence.

> things no other OS platform can. I just got back from visiting what may 
> eventually become the largest stock exchange in the world, and they're 
> new as an OpenVMS customer, moving from HP-UX to OpenVMS 
> disaster-tolerant clusters.


BTW, does anyone have any information on what NASDAQ's intentions are
with regards to its own Tandem based platform versus the  VMS platform
it inherited when it bought OMX ?  I recently saw an interview on BBC
which featured a british employee of NASDAQ and she mentioned the
software being an important asset, but the name she mentioned started
with i (Itrade or something like that) and it didn't seem to be attached
to OMX.


> Even among the engineers from India whom you deprecate are people with 
> 14 years' experience with VMS.

Until someone is allowed to provide numbers, one needs to assume worse
case scenario of a handful of experienced engineers surrounded by a slew
of newbies who will switch jobs before they ramp up their skills on VMS.

The fact that HP has refused to provide any indication of whether the
new staffing levels are equal, higher or lower than the old team leads
one to beleive that it is lower.

> Gartner Group says "With at least three future generations in the 
> Itanium road map (Tukwila, Paulson and Kittson), we expect HP to 
> continue to build and sell Itanium servers for many years to come.

That was an HP sponsored study. Tukwila is EV6. Paulson is EV7 and
Kittson is EV8. Anyone who lived through the murder of Alpha would
understand the comparison.

And the technologies such as Quickpath which were to have given Tukwila
a significant edge ended up implemented on the 8086 first, and giving
8086 1.5 years advance on IA64.



> As a 
> result, OpenVMS (and other OSs) should be fully supported on Itanium 
> servers through 2020 at least.”

Heard the exact same type of promise for Alpha up until June 25 2001.


> And of course there's no technical reason why VMS couldn't be ported to 
> another CPU architecture in the future.

There are now plenty of management reasons against porting VMS beyond IA64.