By Guy Matthews, VNU
Net
The Windows operating system (OS) has taken its first faltering
steps towards a full 64bit future.
Last week, Microsoft chairman Bill Gates proudly showed off a
near complete version of Windows 2000 to run on Intel’s IA-64
hardware, promising that the transition to 64bit would be “much
easier” than the move from 16bit to 32bit.
Tellingly, his words were aimed not at those who will buy or
deploy 64bit Windows, but at code-crunchers attending the Microsoft
Professional Developers Conference in Orlando, Florida. They will
benefit, Gates claimed, from a shiny new instruction set that
Microsoft and Intel have “poured billions into”.
But for everyone else, including Microsoft itself, the move to a
64bit world looks like being anything but simple.
So far, this latest staging post for the Windows OS has been
greeted with relative indifference. This is hardly surprising,
however, as the odd preview version here and there represents
nothing but a pawn in a much longer-term game. But into the future,
Microsoft has a great deal at stake.
Gunning for Unix
The Redmond giant’s clear and stated aim is to drive Windows 2000
further into the enterprise. This poses a threat to the various
incumbent Unix variants that currently dominate the market, the
most obvious being Sun Microsystems’ Solaris.
To this end, Microsoft will initially release two flavours of
64bit Windows, one for servers and one for workstations, which will
be launched simultaneously following the shipment of the first
64bit Intel-based machines later this year.
In yet another attempt to bolster the credibility of its
enterprise ambitions, Microsoft has also generously pre-announced
details of the next two versions of Windows 2000, namely Whistler,
due out in the second half of next year, and Blackcomb, which is
expected a year later.
But there are various flies in this smooth-sounding ointment.
Most notably, Microsoft’s plans have been undermined by its failure
to synchronise shipment of its 64bit baby with the most crucial
part of the jigsaw, Intel’s IA-64 chip.
The Unix community and other Microsoft critics are cock-a-hoop
that almost every flavour of Unix will steal a march on Windows by
being available on IA-64 before it.
IBM’s Monterey/64 will be pretty much ready when Intel launches
the first member of the IA-64 family, Itanium, later this summer.
And doubtless to the particular chagrin of Gates, Linux too
looks like beating Windows to 64bit status.
In fact, some industry watchers argue that a full and final
version of 64bit Windows might not even be available for as much as
six months after the processor arrives.
A lack of 64bit hardware
In an ironic reversal of the early days of Windows application
development, Microsoft has been hobbled by not having access to
fully developed 64bit hardware to test its prototypes on. IBM, Sun
and others, on the other hand, already have their own non-Intel
64bit machines to work with and have had operating systems to match
for some time.
In its defence, Microsoft cites the forthcoming Windows 2000
Datacenter Server, an extension of 32bit Windows 2000, as evidence
that it is already highly active in the enterprise space. It claims
that Datacenter, which is well overdue but scheduled to ship
imminently, is a more than credible offering for demanding
applications such as ecommerce, data mining and online transaction
processing.
The software giant argues that the offering will provide all of
the functionality currently found in Windows 2000 Advanced Server,
but will also include 64Gb of memory, support for 32 processors,
four-node clustering, process control and network load
balancing.
Microsoft’s keenness to stress the strengths of Datacenter
suggests that it is not expecting a stampede for early copies of
64bit Windows. Analysts have certainly warned that no one should
hold their breath waiting for a headlong rush of potential
users.
But does any of this matter in the long term?
Exploiting critical applications
Some analysts say that the real test for 64bit Intel-based systems
in general, and for 64bit Windows in particular, is still some way
off, but will revolve around their ability to exploit the kind of
mission-critical enterprise applications that the architecture was
conceived for. This includes being able to handle the largest of
corporate databases.
Matt Hanrahan, an analyst at Bloor Research, says: “Only a
maniac would want to be first to deploy crucial applications on
64bit technology. If you’re going to use this kind of thing in
anger, it’s got to be totally watertight.”
He believes that the only corporates likely to be rushing for
early editions of 64bit Windows will be those wanting to install it
in a laboratory for test purposes.
However, Hanrahan says that it is still far too early to declare
who the winners and losers in the 64bit space will be, and adds
that Microsoft has as much chance of getting it right or wrong as
anyone else.
Many analysts are describing the early rush by vendors to
produce operating systems that work with Itanium as a phoney war
anyway. Clive Longbottom, an analyst at Strategy Partners, argues:
“For most people, Intel’s McKinley will be the first true 64bit
architecture, but that’s not happening until the latter
half of 2001. Then the race between Microsoft and Linux will really
start.”
If evidence for the irrelevance of Itanium were needed, he
continues, it can be found in the apparent decision by Hewlett
Packard (HP) not to base its servers on IA-64 until McKinley
arrives. HP, as the co-creator of the whole IA-64 project with
Intel, is in a position of some influence and authority in 64bit
matters.
Shock to the systems
Early adopters that feel compelled to move to Itanium – which they
will not be able to swap out for follow-on chips – will find few
system manufacturers in a hurry to develop products for them. And
those systems that do appear will not be cost competitive against
rival systems for an even longer time, claim analysts.
In the same way, application developers are unlikely to bust a
gut to produce the goods for early versions of 64bit Windows.
All this means that Microsoft’s early forays into 64bit
territory are little more than marketing exercises. And the
company may have to resign itself to this situation for some time,
if some influential voices are to be believed. Indeed, analyst firm
Gartner does not foresee that 64bit Windows will find a significant
market presence until 2003.
But if the software giant, on its long trek down the 64bit path,
wants to remind itself why it is bothering at all, it need only
look at the huge sums of money involved in this space. In the first
three months of this year alone, Sun, IBM, Compaq, HP and SGI, each
of which has a 64bit version of Unix that runs on their own 64bit
hardware, claimed the lion’s share of a server market worth
$6.6bn.
However, users would also be unwise to sit out the dance. As
Hanrahan says: “The power that 64bit technology delivers is
something nobody can afford to ignore.”