RE: HP announces new Integrity servers


-----Original Message-----
From: Bill Todd [mailto:billtodd@xxxxxxxxxxxxx]
Sent: September 9, 2006 8:36 PM
To: Info-VAX@xxxxxxxxxxxx
Subject: Re: HP announces new Integrity servers

Main, Kerry wrote:

[snip..]


Sigh .. How many times does it need to be repeated?

Customers are looking for supported, stable and highly available
solutions to run their business.

The ones making the business decisions are not burning up
their meetings
talking about techie chip stuff.

Exactly: they are looking at *standardizing* (for reasons of
cost and
support complexity that have nothing to do with 'techie chip
stuff') on
the *least expensive viable solution*. And beyond any shadow
of a doubt
that's x86-64 in any situation where that solution *is* viable.


Again, you are in the weeds. Customers move to platforms like Windows,
or Linux or OpenVMS for many reasons other than the basic HW. That is
not the issue. When you choose a new std as you call it, you need to
look at the entire HW + OS + App work to move it. The HW is only one
albeit small consideration.

Every platform has strengths and weakneses and challenges. Having stated
this, look at the issues facing someone migrating to a totally new HW
platform + a new OS on your beloved X86 -

Windows - that means .Net which is an OO approach to programming that in
most cases means a total re-write of your current 3GL based application.
Heck, ask anyone who has even tried to move a Windows VB6 application to
Net. Try converting mountains of DCL batch code to something similar in
Windows. Throw away 10-15 years work and start over. Yes, tools can help
but it's the carpenter that builds the house, the tools only help.

Keep in mind re-testing and re-certifying your application will likely
take longer than the actual porting, so 18+ months is likely very light
estimate (if you are lucky enough to find resources that understand both
OpenVMS and Windows code). And of course, in production, it means
continually re-testing your App's with the monthly security patches on
your new platfiorm. And of course, unless you want to re-engineer the
architecture (as opposed to straight port), then get used to running
everything on one system with fail-over clustering only.


That's up to 64 Xeon cores *today* in IBM's X3 line, up to 32
Xeon cores
from Unisys, up to 8 Xeon cores from HP. Up to 16 Opteron
cores *today*
from Sun, up to 8 Opteron cores from HP and IBM). Large
enough in the
case of Xeon for all but the very small niche of those whose needs
exceed 64-core systems (and they'll be able to get more from
x86-64 next
year, when Opteron will start supporting 32-core systems effectively;
IBM's X3 systems may actually support 128-core configurations
as early
as the end of this year if Intel ships quad-core Xeon modules
by then),
and reliable enough for just about anyone who doesn't require *real*
fault-tolerant redundantly-executing hardware.

...

Its probably worth providing this link again for reflection:
http://www.itjungle.com/tug/tug102005-story04.html
"Stop Arguing About Cars and Start Managing Fleets"
What an interesting citation for someone who purports to be a VMS
advocate. That article states clearly, multiple times, and
in multiple
ways, that

"We find ourselves living in a world of parity; at least as
far as Unix,
OS/400, and VMS servers are concerned. What I mean by this
is that as
far as base technology goes, we have a level playing field.
The Windows
and Linux operating systems have made some progress, but
still have to
make up some ground in terms of scalability, availability and
workload
management in order to compete head-to-head with the dominant
Unixes--Solaris, AIX, and HP-UX--or IBM's OS/400-based
iSeries platform
or Hewlett-Packard's OpenVMS platform, which may as well both
be Unix in
terms of its uptime, resiliency, security, and sophistication."


You can selectively cut and micro-analyze what you want, but the point
of the article was all about not getting caught up with the Ghz, GB
techie and OS religion discussions and instead focus on real business
issues like how can IT help the business be more successful?

Think (if you are capable) of what that means: that even as
of today no
one needs VMS at all (given that just-as-good Unix expertise
is so much
easier to come by), that increasingly no one needs anything
but x86-64
(as scalable and reliable solutions on that platform already
exist up to
quite reasonable system sizes - 64 cores from IBM, 16 - 32
cores from
multiple sources for Opteron, and even larger configurations
on the way
for both), and that standardizing on that single platform
(with all its
commodity volume/cost advantages and upon which can run
high-end Unix in
the form or Solaris, dirt-common Windows, and
intermediate-level Linux)
thus has become something of a no-brainer.


The "we don't need mid or high end systems, we can run it all on our
distributed systems!!" low end mantra has been around for the last 20
years. Give it a break.

Yes, these lower end platforms are getting better, but the bar is
constantly being raised as well with other platforms. Customers are
doing massive consolidation projects which requires very highly
available, secure, stable back ends.

Again you keep talking in the HW weeds. Customers need to think about
the whole package (HW, OS, App architecture).

The low end mantra is still very much based on "one-app, one system", so
even if you could run multiple Apps on that high end commodity system
you are talking about, what happens is that you end of having to
virtualize the heck out of to meet that one app, one system culture. As
an eexample, in addition to the technical challenges, try and run 3 or 4
different Windows Bus App's on the same OS and see if the ISV's will
even support it!

Do you really think Customers are not much more concerned about
security, stability and availability than in the past?

Lets look at this from a risk perspective - what Senior Managers do
think about btw.

As opposed to your view, you can also run on a much cheaper HW platform
(IA64) than your current one (Alpha/VAX) with much reduced OS license
costs with much reduced App risks as the OS and App architecture is the
same. If you run multiple App's on the one system, you do not have to
break them up into separate VM's on the target system (see one-app
culture notes above). The new HW also meets all of your scalability
concerns - both up and out. Day to day tasks with this new platform can
be managed with Windows GUI's for junior operations staff not familiar
with the platform. Ref:
http://h71000.www7.hp.com/openvms/products/argus/ (scroll down for
screen shots)
http://h71000.www7.hp.com/openvms/products/availman/index.html

(+ add in Systems Insight Manager + OpenView agents + gui based console
managers + GUI based backups etc). Like all platforms, you still need
more experienced staff doing the set-up for these junior types.

In addition, the new HW can run in the current active-active cluster at
the same time with the old HW to gradually phase out the older HW. No
"big bang" migration required (and big bang migrations scares Mgrs big
time). You also get to keep the same Ap + Ops support staff instead of
having to replace the majority of them. And because you can keep the
existing staff, you do not need to worry that new Operations support
staff will have zero understanding of the way your business actually
works.


Mmm... OpenVMS can certainly hold its own as those running
banks, stock
exchanges, chip manufacturing and major health institutions know.

Not for *new* users, it can't: there's no pool of expertise to draw
from, because virtually all of that expertise has moved
elsewhere save
for that which is still in place (and aging rapidly) from the
days or yore.


See above, with UNIX CLI, Windows Mgmt GUI's, OpenView/SIM type
interfaces + Java / Script based programming styles, it is now much
easier to train those less experienced. And lets not forget that
experienced (not rookies) Windows and UNIX SysAdmins are not that
avaiable either. I know cause I have seen projects trying to find them
as well. Just try and find an experienced Windows resource that can
plan, design, install and integrate a Windows Active Directory project.

Imho, finding a proven, experienced .Net programmer who has real
experience writing highly scalable (SMP aware) .Net code is on par with
finding a senior Cobol resource with Web dev and integration experience.

Not exactly a recipe for growth, or for long-term confidence that its
vendor will continue to do much to keep it even as
competitive as it is
today.


But feel free to keep harping on the techie Mhz and Ghz cool-aid ..

You're the only one trying to divert the conversation in that
direction,
Kerry: I didn't say a word about *anything* technical, just
about the
kinds of high-level management issues that appeal to - well,
high-level
management that makes the final decisions, as the article noted.


I'm not sure what Mgmt types you know, but you are talking about issues
(cores, X86-64 vs IA64 vs Power 5+) that techies talk about. Most senior
Mgmt types would not know the difference between *any* Intel platforms.
To them, an Intel platform is an Intel platform. Or an IBM platform or
an HP platform.


Of course, some people might dispute some of the underlying
assumptions in the above

You think?

The point, as I made clear in the continuation of that sentence that
you've even quoted just below, was that you cited an article which
directly stated that VMS no longer enjoyed any notable advantages as
support for your position. Are you withdrawing that
recommendation now,
or do you just want to pick and choose which portions of the article
people should consider authoritative and which they should ignore?


As stated already - You can selectively cut and micro-analyze what you
want, but the point of the article was all about not getting caught up
with the Ghz, GB techie and OS religion discussions and instead focus on
real business issues like how can IT help the business be more
successful?


, but they follow very directly from combining your own
repetitive droning with the content of the article which you've
recommended we reflect upon. As the adage goes, "Be
careful what you
wish for..."

- bill



[snip..]


And remember that the initial server HW costs are one of
the smallest
components of the overall cost.

Exactly: that's why platforms with dwindling popularity are
generating
dwindling interest, while despite their limitations platforms like
Windows and Linux are growing robustly.


Tell that to almost all the med-large Customers who are doing massive
very aggressive server consolidation projects. The number one target by
far is Windows on x86 platforms. Typical HW consolidation goals ranges
from 20%-40%.

You really should get out more and talk to real med-large Customers.

Between legacy platforms that people choose only when they
have no other
(short-term) choice, and robustly-developed platforms that
people choose
whenever those platforms can meet their needs, which do you really
believe have the better prospects? If the owners of those legacy
platforms had been a lot more aggressive in keeping them current and
competitive, the situation might have been very different -
but that's
water under the bridge at this point.


Give it a break with the "legacy" rhetoric will you? Geeeze, you should
know better.

Microsoft calls Windows 2000 "legacy". Sun calls Solaris 7/8 "legacy".
IBM calls AIX V4 and older "legacy". RH calls its older versions
"legacy".

Are you saying Customers should drop all these platforms?

You know as well as everyone here that every platform has "legacy"
versions. That does not means their current versions should be
considered as legacy.

Btw - I just love the news that Windows Vista is planning to provide the
capability to provide different users with some different levels of
security. What a great new novel concept!!

:-)



Imho, as an example of the importance of benchmarks today

Why do you continue to try to divert the conversation and set
up straw
men? (That was a rhetorical question, of course: anyone who
knows you
at all understands very well why you do that.)


Puuulease, you raised the point about OpenVMS and why it was important
to have a benchmark on the new HW. I was simply responding to that.
Remember this statement you made - " That would be great, since I can't
remember the last time VMS's owners bothered to benchmark anything on
VMS."

When you're on the losing side of an argument, continuing to
bluster for
the benefit of the audience just gives them more opportunity to
understand how full of *** you really are. This is not doing your
employer any favors: just as with the current BoD fiasco, they'd
probably prefer just to minimize all conversation about the
issue - they

And who is now trying to divert the conversation? Leave the BOD
discussion to another thread.

Bill, here is a question for you - have you ever considered taking a
people relationship and communications course? I think you will find
that cursing at people in public places is likely not one of the
recommended best practices for getting respect.

don't need to convince those who already understand and are already
flocking to the *real* 'industry-standard' solutions that HP will
happily sell them, and would just as soon those who don't understand
remain in the dark and continue to pay through the nose for
their legacy
cash cows.


What cool aid have you been drinking?

Based on what you are saying, all mainframe Customers will be soon be
running on x86's as well - right?

Not that there are likely to be all that many people here who still
don't understand, of course: most of those who felt they had
any viable
alternative to VMS have either left already or are waiting for their
current systems to become inadequate while alternatives
continue to mature.

- bill

Yep, they can't wait to jump on this new magic commodity platform that
has all the answers to world hunger.

Even if they have to spend 18-24 high risk filled months getting on to a
new magic platform with a one-app, one system culture that has 5-20
security patches per month, they will love it as they are moving to a HW
platform that is a few $'s less than the alternatives.

Yep, that's what Senior Managers are thinking about alright ...

:-)

Regards

Kerry Main
Senior Consultant
HP Services Canada
Voice: 613-592-4660
Fax: 613-591-4477
kerryDOTmainAThpDOTcom
(remove the DOT's and AT)

OpenVMS - the secure, multi-site OS that just works.
.