x86-64-based Superdomes on the way (says HP)

[John Wallace]

Apologies if this crosses in the post with another posting...

http://www.theregister.co.uk/2011/11/22/hp_to_make_x86_superdomes/
<big snip>
""We are absolutely committed to invest in HP-UX, OpenVMS, and NonStop
operating systems and Itanium platforms," says Bartlett [vice
president of worldwide marketing, strategy, and operations for the BCS
division]. "This announcement is in no way a change in our plans. We
want to retain the best – which is HP-UX and Integrity – and build up
the rest. And that is Linux and Windows on Xeon.""
<big snip>

The original press release can be found at:
http://www.marketwatch.com/story/hp-to-transform-server-market-with-single-platform-for-mission-critical-computing-2011-11-22

[JF Mezei]

A more important snippet from that article:
http://www.theregister.co.uk/2011/11/22/hp_to_make_x86_superdomes/

And in fact, Whitman hinted about it yesterday when participating in her
first conference call with Wall Street analysts to go over HP's Q4
financial results – which, by the way, were not good for BCS, which had
a 23 per cent revenue drop to $535m.

"The BCS business is a declining business. It is a slow decline, but I
don't think you're going to see an accelerating growth rate in that
business," she said. "And so we just have to manage that as best we can
and invest in R&D so we get to a new platform as fast as we possibly can
that allows us to service the clients that need this kind of power."




Notes:

argument that customers are asking for 8086 based superdomes running Linux

argument that BCS is a declining business, and we won't see accelerating
growth rate in that business.

later in the article, the argment that BCS customers don't want a
platform transition and want a stable environment.


The conclusion:

Poulson, yes. Kitson, probably not.


If HP-UX VMS and Tandem customers do want their OS to make it to the
8086, now is the time for those customers to get together and make a
VERY PUBLIC statement that they want HP to port teir OS to x86. Only by
publicly denying Bartlett's "customers don't want it" will HP change its
mind.


As it stands, the death sensence has now been announced for those 3 OSs.



When Tukwila came out, HP didn't bother with performance tests. If
Poulson ends up being as unimpressive compared to whatever 8086 will be
available at the same time, then it will make it quite easy for HP and
Intel toannounce the end of IA64.

In 2004, when LaCarly and Intel announced the 64 bit 8086 with the CSI
(now Quickpath), I had prediced the end of IA64 by 2007 when those would
come out. I was wrong. But now, it appears the end of the line for IA64
is coming.

[MG]

On 22-11-2011 23:24, John Wallace wrote:
> Apologies if this crosses in the post with another posting...
>

> ""We are absolutely committed to invest in HP-UX, OpenVMS, and NonStop
> operating systems and Itanium platforms," says Bartlett [vice
> president of worldwide marketing, strategy, and operations for the BCS
> division]. "This announcement is in no way a change in our plans. We
> want to retain the best – which is HP-UX and Integrity – and build up
> the rest. And that is Linux and Windows on Xeon.""

Who is going to buy an expensive x86-64 Superdome when a cheap 'Dell'
(or whatever) can be had... the main selling point of x86-64, it being
cheap and thus disposable.

- MG

[JF Mezei]

MG wrote:

> Who is going to buy an expensive x86-64 Superdome when a cheap 'Dell'
> (or whatever) can be had... the main selling point of x86-64, it being
> cheap and thus disposable.

When you have a serious mission critical application, you want powerful
hardware that is well built with "carrier grade" hardware and power,
ventilation etc.


Remember that superdomes are now glorified blade enclosures with the
added benefit of shared memory between blades, thanks to quickpath
interconnects via the backpane. (smartly done).


If HP-UX customers can move their oracle apps to Linux on x86, then they
would remain HP customers. And if they can buy a superdome blade
enclosure and populate it with some IA64 to run HP-UX and then load 8086
blades to run Liux, they can port gracefully and keep the investment in
the same superdome blade cabinet.



At this point in time, short of Meg Whitman making a strong pitch for
HP=UX/Tandem/VMS, their demise will be a self fulfilling prophecy.
Remember when HP started to state that VAX-VMS customers wanted a stable
environment and did not want VMS 8.* version as had been promised ?

Whitman jut said that IA64 customers want a stable environment.

If Whitman wants to reverse the downsizing of BCS, one sign would be a
revision of the VMS roadmap to include plans for a next version (as
opposed to fluffy features not assigned any timeframe or version)

Also, the second CEOs start talking about reduction in revenues isntead
of growth potential, it means they have written off that division as
having no future.

Instead of pushing for Hobbyist programme, there should be a push to
open source VMS 8.3, including all the tools necessary to build VMS.

[Michael Kraemer]

JF Mezei schrieb:

>
> If HP-UX VMS and Tandem customers do want their OS to make it to the
> 8086, now is the time for those customers to get together and make a
> VERY PUBLIC statement that they want HP to port teir OS to x86. Only by
> publicly denying Bartlett's "customers don't want it" will HP change its
> mind.

You do know better than HP what their customers want?
What if "those customers" are satisfied with their Itanic
systems, as long as they just stay alive long enough to move
the apps to Linux/x86?

>
> As it stands, the death sensence has now been announced for those 3 OSs.
>

The HP guy said the opposite.

> When Tukwila came out, HP didn't bother with performance tests. If
> Poulson ends up being as unimpressive compared to whatever 8086 will be
> available at the same time, then it will make it quite easy for HP and
> Intel toannounce the end of IA64.

The main performance advantage of Itanic is that it can run
HP-UX, VMS, NSK, seamlessly. No other chip can do that.

> In 2004, when LaCarly and Intel announced the 64 bit 8086 with the CSI
> (now Quickpath), I had prediced the end of IA64 by 2007 when those would
> come out. I was wrong.

By 4+ years.

> But now, it appears the end of the line for IA64
> is coming.

Well, given your "predictive power", the EOL would be long after 2015,
enough time to move apps to other platforms. No need to port.

[JF Mezei]

Michael Kraemer wrote:

> Well, given your "predictive power", the EOL would be long after 2015,
> enough time to move apps to other platforms. No need to port.

So you agree it is the end of the line for HP-UX, Tandem and VMS ?

The message seems to be "start porting to Linux or Windows now, because
a few years from now, new systems will be 8086 only.


If Whitman didn't mean this to be the message then she is utterly
incompetent, just like her predecessors who sent the same message about
BCS and Itanium.

Am listening to the 4th quarter finanial analyst teleconference.
http://h30261.www3.hp.com/phoenix.zhtml?c=71087&p=irol-EventDetails&EventId=3561765


Whitman seems to be much better presenter. She doesn't seem to be
buzzword driven like LaCarly. She does not seem incompetent to me, so I
would conclude that the message we got is the message HP wanted us to get.



For the last year: (in millions of $)

Revenue Earnings
Services 35,954 5,149
ESSN 22,241 3,026
Software 3,217 698
PSG 39,574 2,350
Imaging/Print 25,783 3,973
Financial 3,596 348

ESSN = Enterprise Server, Storage Networking



Within ESSN:

Revenus
8086 Servers 13,531
Storage 4,056
BCS 2,095
Networking 2,569


Note that for "Industry Standard Servers", they now have some Arm based
servers, so it isn't just 8086 anymore.


BCS revenus declined 23% in 12 months, they admit due to Itanium and
Oracle's decision. For 2012, continued decline of BCS is forecasted.

So ESSN is still a big profit centre well aheas of PCs (PSG).



BCS is a declining business, slow decline. Trying to manage the decline.
Still shipping DEC servers and supporting them for 5 years. I think I
heard right. But not sure.


But I heard the following right:
"Get to a new platform as fast as we possbly can."




Whitman admits that they had cut a lot of Muscle out of R&D n previous
years and it will take time to put it back in and get some RPI from the
new investment in R&D.

Overall impression: positive about Whitman being in control/command. Not
much of a "vision", but seems down to earth without bullshit.

But it is pretty clear that BCS has litte future and will be replaced by
industry standard servers where growth is expected.

[MG]

On 23-11-2011 8:06, JF Mezei wrote:
> At this point in time, short of Meg Whitman making a strong pitch for
> HP=UX/Tandem/VMS, their demise will be a self fulfilling prophecy.

It's good to know you at least are able to appreciate all the drama!

> Instead of pushing for Hobbyist programme, there should be a push to
> open source VMS 8.3, including all the tools necessary to build VMS.

That can hardly be any fun, there's no drama involved with that!

- MG

[Phillip Helbig---undress to reply]

In article <4ecc9b8a$0$2205$c3e8da3$4605...@news.astraweb.com>, JF

Mezei <jfmezei...@vaxination.ca> writes:

> Instead of pushing for Hobbyist programme, there should be a push to
> open source VMS 8.3, including all the tools necessary to build VMS.

Apart from the fact that I actually prefer VMS not to be open source (at
least as long as it is available at all from its owner), why in the
world would HP open-source VMS?

[JF Mezei]

Phillip Helbig---undress to reply wrote:

> Apart from the fact that I actually prefer VMS not to be open source (at
> least as long as it is available at all from its owner), why in the
> world would HP open-source VMS?

Once HP announces that VMS will no longer be developped, it can chose to
bury VMS and all its patents, IP, source code etc, or it could let it
out to be open sourced so some of the stuff might survive in the open
source community.

Perhaps Huff and Guy Peleg could get together and start to develop VMS
again from that open source feed :-)


There might also be some contractual obligations to release the source
code should HP stop developping VMS so that thsoe major customers would
be able to fix their own bugs etc.

Having said this, considering that HP still accepts support dollars for
VAX-VMS, it is likely it wll still accept support dollars for Alpha and
IA64 VMS long after they have stopped developping VMS.

[Phillip Helbig---undress to reply]

In article <4ecd861f$0$41491$c3e8da3$fdf4...@news.astraweb.com>, JF

Mezei <jfmezei...@vaxination.ca> writes:

> Phillip Helbig---undress to reply wrote:
>
> > Apart from the fact that I actually prefer VMS not to be open source (at
> > least as long as it is available at all from its owner), why in the
> > world would HP open-source VMS?
>
> Once HP announces that VMS will no longer be developped, it can chose to
> bury VMS and all its patents, IP, source code etc, or it could let it
> out to be open sourced so some of the stuff might survive in the open
> source community.

Yes, of course, and there are other options as well. The question is
WHY.

> Perhaps Huff and Guy Peleg could get together and start to develop VMS
> again from that open source feed :-)

With all due respect, it will take a few more than that.

> There might also be some contractual obligations to release the source
> code should HP stop developping VMS so that thsoe major customers would
> be able to fix their own bugs etc.

I'm not a legal expert, but I am pretty sure that this is not the case.
Has there ever been a similar case where this was the result?

> Having said this, considering that HP still accepts support dollars for
> VAX-VMS, it is likely it wll still accept support dollars for Alpha and
> IA64 VMS long after they have stopped developping VMS.

That is the case for almost everything. The question is how much more
income DEVELOPING VMS will bring in and whether it will offset the costs
of such development.

[JF Mezei]

Phillip Helbig---undress to reply wrote:

>> There might also be some contractual obligations to release the source
>> code should HP stop developping VMS so that thsoe major customers would
>> be able to fix their own bugs etc.
>
> I'm not a legal expert, but I am pretty sure that this is not the case.
> Has there ever been a similar case where this was the result?

Having source code held in trust with the customer gaining access to it
should the vendor cease to exist or cease to develop it is actually
fairly common.

Pretty sure that the military would have such contracts, especially when
you consider the number of changes in ownership with VMS and the
uncertainty about its future since the mid 1990s with many news outlets
declaring VMS dead.

[Michael Kraemer]

JF Mezei schrieb:

> Phillip Helbig---undress to reply wrote:
>
>
>>>There might also be some contractual obligations to release the source
>>>code should HP stop developping VMS so that thsoe major customers would
>>>be able to fix their own bugs etc.
>>
>>I'm not a legal expert, but I am pretty sure that this is not the case.
>>Has there ever been a similar case where this was the result?
>
> Having source code held in trust with the customer gaining access to it
> should the vendor cease to exist or cease to develop it is actually
> fairly common.

For minor apps, maybe, but for an entire OS this is wishful
thinking, I guess. I can't imagine a customer diverting
their scarce resources to offtopic projects like OS development.
The best you can hope here is something similar to
eComStation from Serenity Systems, who desperately try
to give OS/2 an afterlife despite IBM having dropped it.

> Pretty sure that the military would have such contracts, especially when
> you consider the number of changes in ownership with VMS and the
> uncertainty about its future since the mid 1990s with many news outlets
> declaring VMS dead.

Haven't we heard here over and over again how negligible VMS usage
within US military is?

[Michael Kraemer]

JF Mezei schrieb:

> Michael Kraemer wrote:
>
>
>>Well, given your "predictive power", the EOL would be long after 2015,
>>enough time to move apps to other platforms. No need to port.
>
>
> So you agree it is the end of the line for HP-UX, Tandem and VMS ?

Not quite. Just business as usual, reaffirmed.
Those who still want those platforms will get them,
but it's a declining business.

> The message seems to be "start porting to Linux or Windows now, because
> a few years from now, new systems will be 8086 only.

The message is: if you want x86, you'll only get Linux
(or, even worse, Windoze).
So if you want VMS, stop demanding x86.

[JF Mezei]

Michael Kraemer wrote:

> Haven't we heard here over and over again how negligible VMS usage
> within US military is?

It may be negligible, but I suspect that what is left is mission
critical and "embedded" type which means it can't be changed easily and
is there for a long time. Same for some manufacturing plants with VAX
boxes still in production because they are an integral part of the
machinery/automation.

It is a given that VMS isn't going to be pitched for new applications.
And one even wonders if HP-UX will now fall into that one too.

[JF Mezei]

Michael Kraemer wrote:

> Not quite. Just business as usual, reaffirmed.
> Those who still want those platforms will get them,
> but it's a declining business.

Not quite "business as usual" from a PR point of view. Under Livermore,
the message was that they would continue to sell Itaniums and push
HP-UX, while for VMS it was "we'll continue to support existing
customers and offer new hardware". This was putting a positive spin on
things.

Whitman's message in that teleconference was rather blunt: BCS is
shrinking by 23%, given up on it with the only effort being to try to
control the rate of decline. She added that they would make an effort
to find a new platform. So the hints are much much stronger than before.


Oracle's decision to desupport IA64 may have wounded BCS much more than
we may have tought. It basically relegates HP-UX to the same status as
VMS: no expectation of new customers so just a question of taking care
of the remaining installed base.


had Pfeiffer been replaced by a real CEO at Compaq, it is not clear
how/when/if Alpha would have been replaced by IA64.

Because Curly was in a hurry to find a buyer for Compaq (since he wasn't
competent enough to lead the company back to success), he had no choice
but to prematurely announce the murder of Alpha without any warning
because that may have been a precondition set by LaCarly. This was
despite Alpha being a great chip with great future of tech innovations
planned for it.

In the case of HP, they have been much smarter at controlling
expectations on IA64 and by the time HP formally announces the end of
IA64, customers will be happy to finally put an end to this long story
where IA64 has always played catch up, never any true technological edge
and thus no real asset to anyone.

And IA64 will have had a lifetime of greater than 10 years, despite its
death having been predicted since the days of Merced. So there won't be
any accusations of HP killing a chip with a bright future or not having
given IA64 a chance.

Consider that if BCS sales drop by 20-25% per year, HP won't be able to
justify paying for continued development of IA64 for much longer.
Poulson will come out, but I am not sure about its succesor. By the
time the successor would come out, the market may be too small to pay
for its development costs.

> The message is: if you want x86, you'll only get Linux
> (or, even worse, Windoze).
> So if you want VMS, stop demanding x86.

The underlying message from Whitman: "if you really need your OS (HP-UX,
NSK, VMS) to continue beyond IA64, talk to us NOW so we can gauge demand
for a port because right now, our thoughts is that those systems won't
be ported beyond IA64."

I expects important customers like NASDAQ will have private (off the
record) dinners at Whitman's residence to discuss this and this will
happen quickly (or has already happened).

[Michael Kraemer]

JF Mezei schrieb:

> It may be negligible, but I suspect that what is left is mission
> critical and "embedded" type which means it can't be changed easily and
> is there for a long time. Same for some manufacturing plants with VAX
> boxes still in production because they are an integral part of the
> machinery/automation.

But for this one doesn't need to port off the Itanic.
Those systems just keep running until the hardware
it controls is replaced by something else.
If they're still on VAX and thus on an OS
whose development has stopped ten years ago,
they will even survive if HP stops selling
VMS tomorrow.

[JF Mezei]

Michael Kraemer wrote:

> But for this one doesn't need to port off the Itanic.
> Those systems just keep running until the hardware
> it controls is replaced by something else.
> If they're still on VAX and thus on an OS
> whose development has stopped ten years ago,
> they will even survive if HP stops selling
> VMS tomorrow.

However, a formal end to VMS development would send the message that HP
is getting out of that business and thus generate questions about
continued support an availability of spare parts etc. It could be that
HP would provide sufficient assurance and confidence that they will
continue to support them. Or maybe not.

For customers on Alpha and IA64 however, the story is different because
they still expect software updates and new VMS versions, so formal end
to VMS development will affect them. (but so is the formal end of Oracle
and RDB development as per Oracle's decision to abandon IA64).

[Jan-Erik Soderholm]

> and RDB development...

It's perfectly clear that you have no idea whatsoever what'you are talking
about. Have you been to any of the Rdb Tech update events lately ?
There is *no* change from the Rdb group in regard to the Oracle decission.
Apart from one (minor) thing, the version number is frozen at 7.3.x.
But that is not planned to change the actual content of Rdb.

> as per Oracle's decision to abandon IA64).

Note that it is "only" the server parts that are affected, the client
kits are planned to contionue to be developed to support future
Oracle Server versions. So from an application point of view, the
code will continue to work as before, but the *server* will have to
run on some other box. As far as I understand...

[Paul Sture]

On Fri, 25 Nov 2011 00:06:03 -0500, JF Mezei wrote:

> Not quite "business as usual" from a PR point of view

It's been a pretty poor time for PR. On top of the kerfuffles with
Oracle, the shall we, shan't we? bit with the PCs and tablet cannot have
inspired confidence. Corporate customers look for stability in their
suppliers.

--
Paul Sture

[Michael Kraemer]

In article <janpc7$o8m$1...@news.albasani.net>, Jan-Erik Soderholm

... VMS is only a server OS, so stopping development of server
apps is a major blow.

[Richard B. Gilbert]

"News outlets????" Which ones do you have in mind?

As for "ownership", I think that's DEC, Compaq, and H-P. It doesn't
seem excessive!

[JF Mezei]

Richard B. Gilbert wrote:

> "News outlets????" Which ones do you have in mind?

Oh come on now. Gartner and many IT newsmedia kept saying VMS was
dead/dying or soon to be dead troughout the late 1990s. After Compaq
killed Alpha the media stopped bothering proclaiming the obvious.

> As for "ownership", I think that's DEC, Compaq, and H-P. It doesn't
> seem excessive!

VMS and NSK have changed hands moe often than other mainstream operating
systems. VMS especially since there was so little time between Compaq
buying Digital and HP buying Compaq.

[MG]

On 25-11-2011 4:26, Michael Kraemer wrote:
^^^^

> Haven't we heard here over and over again how negligible VMS usage
> within US military is?

I see you rose and shone extra early, for your cause. Now that's
dedication, I must admit.

- MG

[Michael Kraemer]

MG schrieb:

Not quite.
To fall asleep, one needs some boring, repetitive stuff,
like this "DoD will rescue VMS, for sure" urban legend.
Just counting sheep.

[JF Mezei]

Michael Kraemer wrote:

> Not quite.
> To fall asleep, one needs some boring, repetitive stuff,
> like this "DoD will rescue VMS, for sure" urban legend.
> Just counting sheep.

Nobody is stating that DoD will rescue VMS.

But if they still have some legacy stuff running on VMS, they would want
some sort of assurance that there will be support until they can migrate
away from it.

It seems clear that the amount of VMS at the USA military has gone way
down. But there *may* be some misssion critical stuff left.

[MG]

On 26-11-2011 4:19, Michael Kraemer wrote:
> Not quite.
> To fall asleep, one needs some boring, repetitive stuff,
> like this "DoD will rescue VMS, for sure" urban legend.
> Just counting sheep.

You must have the luxury of a late afternoon job then?

- MG

[MG]

By the way...

On 26-11-2011 4:19, Michael Kraemer wrote:

^^^^

LOL.

- MG

[Phillip Helbig---undress to reply]

In article <janpc7$o8m$1...@news.albasani.net>, Jan-Erik Soderholm
<jan-erik....@telia.com> writes:

> It's perfectly clear that you have no idea whatsoever what'you are talking
> about. Have you been to any of the Rdb Tech update events lately ?
> There is *no* change from the Rdb group in regard to the Oracle decission.
> Apart from one (minor) thing, the version number is frozen at 7.3.x.
> But that is not planned to change the actual content of Rdb.

Yes, the Rdb folks said that they will conform to the letter of
Ellison's decree in that they will never release 7.4, but rather 7.3.x,
where x can become arbitrarily large. How much of this gets through to
the decision makers is another question. (To be sure, Rdb did this
before, when what should have been 8.0 was branded 7.2 or whatever.)
However, it is clear that Ellison (perhaps because he now owns Sun
hardware) is not as sympathetic to other hardware as he was before. A
side effect of this is that it looks worse for VMS.

> > as per Oracle's decision to abandon IA64).
>

> Note that it is "only" the server parts that are affected, the client
> kits are planned to contionue to be developed to support future
> Oracle Server versions. So from an application point of view, the
> code will continue to work as before, but the *server* will have to
> run on some other box. As far as I understand...

OK, but that rules out VMS.

[Phillip Helbig---undress to reply]

In article <4ed05e29$0$19600$c3e8da3$c8b7...@news.astraweb.com>, JF

Mezei <jfmezei...@vaxination.ca> writes:

> Nobody is stating that DoD will rescue VMS.
>
> But if they still have some legacy stuff running on VMS, they would want
> some sort of assurance that there will be support until they can migrate
> away from it.

At best, it means some sort of support FOR THEM, not for other customers
(and certainly no patch access for hobbyists).

[Michael Kraemer]

JF Mezei schrieb:

> Michael Kraemer wrote:
>
>
>>Not quite.
>>To fall asleep, one needs some boring, repetitive stuff,
>>like this "DoD will rescue VMS, for sure" urban legend.
>>Just counting sheep.
>
>
>
> Nobody is stating that DoD will rescue VMS.

Quite a few people do.
Just the other day I read someone claiming
that HP for sure would port VMS to x86
"because of those long term contracts for DoD".
And this although HP has clearly stated in that SD2
announcement that it aint going to happen.
Urban legends are hard to kill it seems.

> But if they still have some legacy stuff running on VMS, they would want
> some sort of assurance that there will be support until they can migrate
> away from it.

Well, if it ran last year, it would also run next year,
even if HP stopped selling it.
The real critical point is ISV support going away.

> It seems clear that the amount of VMS at the USA military has gone way
> down. But there *may* be some misssion critical stuff left.

This may well be, but associated support contracts just preserve the
past and have little meaning for VMS' future. A support contract
is not a "development contract".
Even if special contracts with DoD existed, they would mean
zilch for the viability of VMS as a general purpose OS for
the rest of the world.

[VAXman-]

In article <jaqr41$9uk$1...@weretis.net>, Michael Kraemer <M.Kr...@gsi.de> writes:
>JF Mezei schrieb:

>> Michael Kraemer wrote:
>>
>>
>>>Not quite.
>>>To fall asleep, one needs some boring, repetitive stuff,
>>>like this "DoD will rescue VMS, for sure" urban legend.
>>>Just counting sheep.
>>
>>
>>
>> Nobody is stating that DoD will rescue VMS.
>

>Quite a few people do.
>Just the other day I read someone claiming
>that HP for sure would port VMS to x86
>"because of those long term contracts for DoD".

Reference?

--
VAXman- A Bored Certified VMS Kernel Mode Hacker VAXman(at)TMESIS(dot)ORG

All your spirit rack abuses, come to haunt you back by day.
All your Byzantine excuses, given time, given you away.

[JF Mezei]

Michael Kraemer wrote:

> This may well be, but associated support contracts just preserve the
> past and have little meaning for VMS' future. A support contract
> is not a "development contract".

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.

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.


On the other side of the coin, perhaps Whitman sent this trial balloon
out to see if the HP-UX/VMS/NSK customer base will revolt and demand
their OS be ported to 8086 servers 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.


In fairness to HP, they have upheld their word to continue to support
existing customers and give them time to migrate to a new platform (as
per Stallard's original May 7th 2002 memo). And they will do this for
Alpha and IA64 customers.

However, those customers who need new features and support for new
interconnects (such a disk arrays, IP stack features etc) will find that
"support only" will not be sufficient and they will need to migrate to
an actively developped platform eventually.

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

[Phillip Helbig---undress to reply]

In article <4ed1662c$0$2234$c3e8da3$76a7...@news.astraweb.com>, JF

Mezei <jfmezei...@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.

[Richard B. Gilbert]

Who is going to pay him? Who is going to pay for the necessary
anti-nausea medication?

[MG]

On 27-11-2011 11:26, Phillip Helbig---undress to reply wrote:
> 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.

Way more than 'just' 20 by now, but yes.

Every time I tell someone about VMS, they're the vast majority of the
time (read: ~99%) interested to try VMS. The problem is, most people
simply haven't heard of VMS. Like today, when I was at university, I
told someone about my remote access cluster initiative and about the
"Deathrow" cluster. The person was instantly interested, fascinated
that 'there's more than merely Windows and UNIX/-derivatives!' (A
typical response that I usually get, when I bring this things up.)

The way I see it: we're in the Kali Yuga (VMS Engineering will probably
understand what I mean) of IT and computing in general.

- MG

[Phillip Helbig---undress to reply]

In article <4ed3dea8$0$6909$e4fe...@news2.news.xs4all.nl>, MG
<marc...@SPAMxs4all.nl> writes:

> On 27-11-2011 11:26, Phillip Helbig---undress to reply wrote:
> > 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.
>

> Way more than 'just' 20 by now, but yes.

Not that much. When was clustering introduced? Late 1980's?

[MG]

VAXcluster, 1984; so (27), closer to 30 than 20.

- MG

[John Wallace]

On Nov 29, 8:14 pm, hel...@astro.multiCLOTHESvax.de (Phillip Helbig---
undress to reply) wrote:
> In article <4ed3dea8$0$6909$e4fe5...@news2.news.xs4all.nl>, MG

>
> <marcog...@SPAMxs4all.nl> writes:
> > On 27-11-2011 11:26, Phillip Helbig---undress to reply wrote:
> > > 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.
>
> > Way more than 'just' 20 by now, but yes.
>
> Not that much.  When was clustering introduced?  Late 1980's?

I'm not aware of any other system that does what VMS does, but one
which was briefly interesting in the 1990s was briefly owned by CPQ
(having been developed for Tandem) and eventually became open source:
the former Compaq/Tandem NonStop Clusters for UNIXware/SCO UNIX. The
open source successor is OpenSSI but it doesn't seem to be very
visible or very actively developed, which seems to me (as an observer
rather than a user) be a bit of a shame.

A great many features recognisable to VMS cluster folks would be
recognisable here: single system image, single clusterwide filesystem
(and given that pretty much everything on UNIX is a file or accessible
as a file, think what that means: clusterwide processes, etc). A
little bit more can be read at:
http://en.wikipedia.org/wiki/UnixWare_NonStop_Clusters and elsewhere.

It's UNIX, so it's not VMS. But it's clustered UNIX.

[Ken Robinson]

I worked on my first VAX Cluster in late 1984. In fact the cluster I was managing was the first one that our local DEC office installed and they were stymied for two weeks by a cryptic error message that occurred when they tried to boot the 3 systems with separate system disks into the cluster for the first time.

Ken

[Paul Sture]

I recall turning down a job in 1986 because they weren't intending to use
clusters. As things turned out, I didn't manage to get my hands on one
(in the management sense) until 1991.



--
Paul Sture

[Richard B. Gilbert]

Turning down a job? Talk about nostalgia!!!!