No subject

Michael Turner leap@REDACTED
Sat Aug 15 07:08:33 CEST 2009

On 8/15/2009, "Benjamin Tolputt" <btolputt@REDACTED> wrote:

>Dave Pawson wrote:
>> I think Microsoft also sell servers?
>> More assumptions
>Indeed they do. But in terms of their usage in high-throughput server
>applications, their deployment is not the majority case. I am not trying
>to start an operating system flame-war here... hell, I am primarily a
>Windows developer.

True, but also, I believe, somewhat beside the point.  If I may

Eons ago (i.e., during the Internet Bubble), I did something that an
employee should never, ever, do: I contradicted my boss during an
important sales pitch.

My boss was selling a LAMP stack with some e-Commerce thing on top.  The
customer really only wanted the e-Commerce.  My boss kept flogging
(rather ignorantly) its LAMP infrastructure instead.

The guy enduring all this techno-gibberish asked my boss: Why does it
have to be Linux?  My boss answered: because you need Apache to be on
the Web, and Apache doesn't run on Windows NT.

Suffering in silence up to that point, I cleared my throat and corrected
him: Apache had, in fact, been running on NT since some fairly early
releases.  MySql and PHP also ran on NT.

Because my boss turned to me then and frowned at my sinfulness, he
didn't see what I noticed at the same moment: the relief washing over
the face of the customer.

The sale was made.  It was understood that the initial delivered system,
which was supposedly urgently required, would be running Linux.  But
only as a stopgap.

The plan was to switch to Windows NT later, in due time.

But they never did.

Apache's portability to Windows NT got a Linux system through the door,
where it proved itself on its merits.  I doubt my story is an isolated
case -- we were as generic as you could get, for web startups back then.
 And the client was a pretty generic client.

I also have little doubt that Apache's availability on NT got some
NT-based IT departments on the Web, which put those IT departments
(hitherto mainly cost centers) solidly astride new revenue channels for
the companies they served.  Revenue channels that would be endangered if
NT crashed.  Which NT did -- rather a lot, in fact.  And in that way,
the portability of LAMP-minus-the-L to NT platforms helped sell even
more Linux and BSD into that new arena of demand.

Many more eons ago, there was a young man who committed heresy: he was
writing a version of Unix (yay!) that would be GPLed (yay!), but (unlike
GNU software) was targeted *initially* for only one hardware system
architecture: IBM PC compatibles (BOO!).  This was an ugly, highly
sub-optimal architecture that had become standard in large part because
the OSes and apps for it were very widely adopted de facto standards,
through an evolutionary path going back to Z-80s, the S-100 bus, and
CP/M (close your eyes, children).  The system software on most computers
was eventually almost entirely from Microsoft.  And so were many of the
apps.  And this Macdonalds-like predictability (albeit with
Macdonalds-like low quality) in computer software products somehow sold
software like cheeseburgers, which helped beat down the cost of hardware
to the point where a Finnish college student could hack on operating
systems at home without even needing a continuous dialup connection
after a certain point.  What, you don't see anything miraculous about
that?  Bah.  Kids these days.  So spoiled.

This young man has since become legend.  His heresy, however -- Thou
Shalt Not Worship Beautiful Universal Portability If You'll Get to
There Faster by Focusing on Some Cheap and Ugly Single-Platform Route --
seems to have been largely forgotten.

But it's true: Wintel helped make Linux possible.  If it hadn't been
Wintel, it might have been some even more revolting oligopoly.

So, to get back to the point: yes, Microsoft lags on the server side, and
it probably always will.  But there are still big companies out there
populated by people who want to run software from other big companies,
because . . . because . . .  I mean, because they are big, and, y'know,
everybody buys from them, y'know?  I.e., companies where tech buy
decisions are still made by people who arguably shouldn't be making

Every corporate IT generation will have its own "Nobody ever got fired
for buying ___".  When I started programming, the blank was filled by

That guy in that meeting where I saved the sale?  Maybe even *he* knew
Linux would be the better bet.  But if so, I doubt his boss did, and
maybe that boss could never have set aside enough time to be persuaded
of why his prejudices were wrong.  And maybe that boss never even
discovered that he'd gotten Linux for the long run, instead of NT. 
More likely, that boss *did* discover it, later, but took credit for
making an insanely great decision after it had proved to be, at least, a
reasonably good one.

I hate this kind of thing.  Maybe we're inexorably leaving that world
behind.  But I don't think we're out of the woods yet.  And Erlang
straggles through those woods just like everything else has to.

Go ahead, whip me, beat me, call me a cynical marketing flack.  I've
learned to love it.

-michael turner

More information about the erlang-questions mailing list