Sun Microsystems has spent the past half dozen years
attacking Microsoft and launching court cases against it. But if
Sun did go under, no one would have any doubt which companies would
be most responsible for its demise. The main one would, of course,
be Intel. Its many assistants would be led by Dell, Red Hat, SuSE
and the programmers who contribute to the GNU/Linux operating
system.
For behind all the rhetoric, Sun's main business is selling servers
with Sun Sparc processors running Sun's Solaris operating system.
And if Sun's customers defect, or it fails to win new accounts,
there is no doubt where the bulk of the lost business will go: to
cheaper Intel servers running Linux.
At the low end, the companies most likely to hurt Sun are Dell and
Hewlett-Packard. At the high end, they are Hewlett-Packard, IBM and
rack system suppliers such as Rackable Systems, whose customers
include Yahoo and Google.
Sun must have seen this coming. Dell and other Wintel suppliers
have already taken over the workstation market that used to be
dominated by Risc/Unix companies such as Sun and Silicon Graphics.
They have also taken over the bulk of the entry-level server
business with Windows servers, and the web server business with
Linux/Apache/MySQL/ PHP (Lamp) boxes. Now Intel systems are
starting to arrive at the high end.
The current record-holder for the number of TPC-C transactions is
an HP Superdome running Windows Server 2003 on Intel Itanium 2
processors - that is for raw, non-clustered performance, not just
price/performance. Linux cannot match that at the moment, but it
will.
It is hard to compete with Wintel/Lintel economies of scale. In
2002, Dell's turnover was $34.5bn (£20.7bn), up from $31.2bn, and
its gross profit margin was a razor-thin 17.9%. Sun's annual
turnover was $12.5bn in 2002, down from $18.3bn, and its gross
margin on products and services was 44.6%. Both companies spend
roughly the same on R&D - about $450m a quarter - but Dell does
not have Sparc, Solaris or Java to support. Who would you back to
win?
In the late 1980s, I thought Sun's bet on Risc chips and Unix
looked a very good one. We might all be better off if it had been
right. By 1995, however, it was dubious and by 1997 it was clearly
wrong. Sun has had six years to do something about it. Did it help
to spend that time bashing Microsoft instead of focusing on the
real threat from Intel and Dell?
Jack Schofield is computer editor of the Guardian