Conference 7.286::digital

>    would be able to use it. According to Gene Austin, Compaq Systems
>    Division VP of Marketing, the unmodified DEC stuff only works with
>    Alpha boxes and bringing it to other NT platforms would have required
>    heavy re-engineering. Also, the DEC scheme is software-only and

     Wow - what a load of FUD. Digital's vision of NT clusters from
     it's inception and demoed around the world was for a totally
     open technology - from the server to the interconnect to the
     storage - absolutely no dependencies on any vendor.

     Recently, MS has been courting other vendors (See note 4195) to
     continue to develop and foster NT clusters in the industry. I
     don't know how this plays in to our licensing agreement but it
     appears that MS has adopted our underlying technology. 

     It seems that Digital will do anything to ensure that NT clusters 
     gets into the enterprise thus positioning MCS to be the vendor of 
     choice for service (note the deal with compaq) and Digital could
     promote Alpha servers as the best server platform (price/performance)
     for NT clusters.

     Anyway, that's my cut.

    when you crawl into bed with someone, have a good time, and then, as
    the sun begins to light the day and you get a look at your bed-partner,
    things take on a whole new perspective.
    
    tony

RE: .2

	Yea, you begin to wonder exactly HOW much you drank the night
	before and you promise yourself you won't do it again and
	sure enough, you do, usually about 6 months later.

	:)

							mike

Having been around Compaq people quite a bit at a stint at Microsoft:

Compaq complains about every piece of technology that 'was not invented here'.
I once sat through a DLT presentation, after 5 years of toting TK50's, 70's
etc. in my brief case, and had to listen to how Compaq was the almighty DLT
vendor because they discovered a bug in the firmware, thereby saving the
industry from DEC...and they get that trait from MS if ya ask me...

MS is our partner and everything, but I tell you one thing that I'm sick of,
is Engineers thinking that everything they touch has to be 're-wired', they're
like the Tim Allen of software half the time.  I can't tell you the headaches
I've experienced over WINS and DHCP, the hours and hours trying to get that
'rewired' IP integrated with 'real' UNIX/VMS systems that talk normal IP.
But NNNNNNNOOOOOOOOOOOOOO....MS had to rewire IP...

Whatever happens, we get a piece of the pie, so I say let the kids play....let
'em FUD us till the cows come home...one thing I've learned in the past six
months or so is that MS/Compaq like to think they are in the midst of some
kind of revolution, and they are the driving force.  To a certain extent,
they have a point, but geez, the hype is so thick at times my hip waders
need a serious extension.

Some of the non-disclosures I've sat in in the past with Compaq were so thick
with 'Compaq and MS are gonna rule the world' hype, I had to excuse myself
to go get sick.

-Stephen

    If we let the kids play they will eat our lunch. Our problem has
    always been not jumping into or creating the hype that you are sick
    of.  If they hype it up as a revolution even though we know better the
    market will think it it is a revolution. Those with the best hype will win.
                                      

    Compaq/MS know that if you build a better path to your door the world
    will buy your mousetraps.
    
    Steve

     get a big stick, and hit those kids real hard. Often. Until they
    understand. 
    
    
    [..oops... sorry.. this not allowed anymore...].
    
    Microsquich know almost *nothing* about enterprise compute. And their
    customers (*big* ones) are beginning to see thru the emperors new/old
    clothes. Compaq is just extra fog... they'll get found out too.
    
    However, if we have a horn, let's blow on it..
    
    pick up that big stick.
    
    
    
    AW

> MS is our partner and everything, but I tell you one thing that I'm sick of,
> is Engineers thinking that everything they touch has to be 're-wired', they're
> like the Tim Allen of software half the time.  I can't tell you the headaches
> I've experienced over WINS and DHCP, the hours and hours trying to get that
> 'rewired' IP integrated with 'real' UNIX/VMS systems that talk normal IP.
> But NNNNNNNOOOOOOOOOOOOOO....MS had to rewire IP...

MS is not the only one to attempt to needlessly re-engineer software.
VMS engineering did the same thing with X11R1 server code from MIT back in
1987-89.  Figuring they could end up improving the performance of the
MIT code, they tweaked it and by Field Test had it running at 25% the
performance of the public-domain software.  By FCS time, I think it was
upto around 60% of the performance they would have achieved without
re-engineering.

UNIX engineering did the same thing with the initial integration of NFS
into ULTRIX (I believe it was in V2.0).  There was a gratuitous dumping
of SUN's VFS file-system-switch for a home-grown GFS, causing the ULTRIX
release to slip several months while they debugged it.  Then they presented
a paper on the work at a USENIX conference I happened to attend.  One of
the questions asked at the end of the talk was "why did you bother to
invent GFS, when VFS does the job fine?"  I still remember the reply given
by the Digital engineer in front of an audience of several hundred people:
"Because we are engineers and we get paid to write code".

As competitive pressure heats up, and profitability becomes more of an
issue, engineering for the sake of engineering, gives way to engineering
to solve customer problems.  I think we at Digital, have experienced this.
Microsoft hasn't felt the heat yet!

    Similar article in dec computing 1/10/95
    
    Mostly Compaq blowing its trumpet. eg:-
    
    Move to the big league.
    
    "'We may or may not use some of it [DEC's NT clustering] in the final
    solution. Digital's technology only runs on Digital hardware and does
    not have the level of abstraction that would make it run on Compaq'"
    
    Bob Muglia, VP business systems division, Microsoft
    
    It goes on and Tandem get a mention.
    
    Was our NT cluster stuff Digital specific?
    
    Andrew

    NO, our stuff was NOT Digital specific. 
    
    BUT our SUPPORT was /  will be / may be Digital specific. Engineers
    need to test on different hardware platforms. The hardware platforms
    chosen to test on are our own (we need an edge somehow!), both Intel
    (Prioris) and Alpha (2100).
    
    Cheers, Chris

    Today, for time to market reasons there may soon(ish) be TWO sets of NT
    clustering products using Digital technology. Digital's will be first
    to market, and will be qualified on, and supported on, Digital hardware
    (some Intel and some Alpha, as Chris says).
    
    Meanwhile, MS have access to our technology and will in due course
    bring a product to market. Long-term, the two will merge again.
    
    I personally have a bit of a problem with the words "Microsoft",
    "qualify", and "support" in the same sentence, as will a large number
    of today's MS customers who understand that the needs of the DP
    environment aren't identical with the desktop. So, we have some
    advantage up front on this subject, we need to make the most of it.
    
    Source for facts in paragraph 1 above: LJSRV1::ALT1:[NT_CLSTR],
    specifically, clusters.ppt. Have a look, it looked good when I looked.
    
    Compaq have a big name and have a good positioning in the PC server
    market today. For a lot of customers PC server = DP = Compaq. There, we
    have a battle to win.
    
    regards
    john