TCLUG Archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Can Linux avoid Microsoft's NT trap?
An interesting article:
http://www.theregister.co.uk/990514-000025.html
Posted 14/05/99 2:02pm by John Lettice
Can Linux avoid Microsoft's NT trap?
Microsoft's spinmeisters are no doubt highly satisfied with the
progress of their
assault on the Linux community so far this week. The 'put up or
shut up' challenge is of
course a trap - but yes, what is the Linux world going to do about
it?
Discussions of this so far seem pretty rational and realistic. It's
generally accepted
that Linux will perform better on low-end systems, so the
Microsoft-Mindcraft offer of
testing at this end is welcome. But it's also accepted that the PC
Week Labs-hosted
tests proposed will quite possibly show NT beating Linux on bigger
systems. If it did
so, that would allow Microsoft and Mindcraft, whose previous study
came under heavy
Linux fire, to say they'd been right all along.
So what do you do? If you accept the challenge you'll quite
possibly lose, and if you
duck it you lose too. Microsoft has already accused the Linux
community of being
"slow to respond," so if nobody accepts on Linux's behalf (as has
been pointed out,
the notion of a community responding is pretty weird), Microsoft
will just scream this
louder.
So far so good for the spinmeisters, but let's think for a moment
about the narrowness
of the battlefield, and who's choosing it. It's being pitched by
Microsoft as a shootout
between Linux and NT, with greater emphasis on scalability than is
currently
convenient for Linux. Other systems beat NT in scalability (that
Ray Anderson of SCO
mailed us just this morning boasting about UnixWare 7 doing this),
but if you don't take
them into account, and fight on ground favourable to NT, then you
get a result that says
"NT is the clear winner!"
This is not rocket science. The scum of the software business
(i.e., practically all of
them except Ray) set up their benchmarks to show their products in
the best possible
light against the competition. They can set the tests up brutally,
tuning one installation
and not bothering too much about another, and this is what the
Great and Good of the
Linux world say Mindcraft did for its first, disputed report. But
even if they don't do that,
they can still choose the 'level' playing field that best suits
their product. And why not?
You're not going to pay for lab tests that show your product
stinks, are you? Well, not
deliberately, anyway.
Think about how this works in the commercial world, and the
possibility of a viable
answer to the Microsoft challenge begins to emerge. Microsoft has
chosen the
territory it wants to fight on, but if it had challenged, say,
Oracle or Sun, how would they
respond? They would of course point out that the shoot-out was
being planned
deliberately for ground that favoured Microsoft, and fire back with
alternative
conditions and their own 'totally impartial' test data that showed
how great their stuff
was instead.
But no, Linux doesn't have to be that despicable. Linux is likely
to perform better and
more cost-effectively at the low-end, so there has to be more
weight insisted on here.
As currently proposed by Microcraft, the low-end aspect looks like
tokenism. Then
there's cost of ownership, uptime (Microsoft's dangerous references
to how good at
this NT is have been remarked on elsewhere, and could be exploited)
and so on.
There's the whole artificiality of tests of this sort too - Linux
does well in the real world,
so some real world factors ought to be brought in.
And instead of agreeing to Microsoft's narrowing of the field, why
not broaden it out
drastically? Why not say, 'look, if you want to prove NT's so
great, why don't we turn
this into a proper tests that shows the good points and bad points
of all of the players?
Why don't you let Novell, Sun, Oracle and SCO play too?' Put up or
shut up?