The problem at Sun isn't at the top

Last week's Economist devoted a full page to a thinly veiled personal attack on Scott McNealy. As usual in cases like this the author is full of absolute certainties:

during the dotcom boom of the late 1990s, Sun was the preferred maker of the powerful server and network computers that new-economy companies bought in bulk with the money that venture capitalists were throwing at them. Mr McNealy thus became an emblem of the internet boom. He captured the Zeitgeist with two great slogans, proclaiming that the network is the computer and boasting that Sun (as the supplier of the machines that ran the internet) was the dot in dotcom.

Then the boom turned to bust, and suddenly hardly anybody wanted to buy computers, least of all those made by Sun. That was because Sun had been stubbornly ignoring an industry trend toward low-cost, standardised hardware and software that customers could buy piecemeal and patch together. Sun's rivals, such as Hewlett-Packard, IBM, Dell, Intel and Microsoft had embraced this trend and happily made whatever pieces they were good at. But Mr McNealy, who was one of four co-founders of the company in 1982 and became chief executive at the age of 29, insisted on a complete, proprietary Sun solution based around Sun's own processor chips (called Sparc) and operating-system software (called Solaris). It was an all-or-nothing proposition for customers. Many customers chose nothing.

As usual, however, these certainties are simply and completely wrong. What generally happened during the dot dumb bust was that many people bought Windows NT, discovered it wouldn't work, and jumped on the Sun bandwagon without changing IT staffs or business models while a lot of big data processing shops skipped the Windows NT step but made the same management mistakes - and the combination ultimately stuck Sun with hundreds of millions in bad receivables, a reputation associated with web failures, and a wildly competitive market selling Sun boxes out of receivership at a few cents on the dollar.

Sun's response was to invest in R&D while the market settled - and today they virtually own the high end Opteron market, have the first products from their next generation architecture in the market, and have established the world's fastest growing open source community around Solaris.

Of course that isn't what the Economist thinks happened. Their reporter repeats the bromide "Sun, not Microsoft, has become the clear loser from the rise of Linux, the open-source operating system, which many customers are using to replace their old Sun systems" and goes on to explain Sun's x86 commitment as the result of of expediency in the face of dispair:

Gradually, Mr McNealy came around to the new way of doing things. Always a good sport, he mounted a stage in 2002, dressed as a penguin (the mascot of Linux) to announce that he would offer Linux on Sun servers. He struck a deal with AMD, a chipmaker that is the main rival to Intel, so that Sun could supply computers built around AMD's chips in addition to its own Sparc chips. Mr McNealy bought a company which had been started by Andy Bechtolsheim, another of Sun's four founders, that made computers based on industry standards, the antithesis of Sun's approach. He bought a large data-storage company, too.

Look at this comment carefully and what shines through is that writer knows some things for sure - and expects you to know them too. For him a move to Linux is a move away from Unix, for him x86 defines "industry standard" but standards compliant software doesn't even exist as an issue, to him Sun's Opteron products are an attempt to get in on a burgeoning market developed by Sun's competitors.

He's wrong on all counts.

On the Linux front the point is simply that Linux is Unix, so switching from Solaris on older SPARC boxes to Linux on newer Dell or other x86 servers is trivially easy while a switch from Windows 2000 Server to Linux generally only works in the longer term if accompanied by staff change.

The x86/Opteron issue is wholly different. Back in 1999 and 2000 the big threat to Sun wasn't Linux on x86 - something most Sun people saw then (and still see today) as positive for all of Unix and therefore for Sun - but Windows on Itanium. At the time the Itanium seemed to have the potential to successfully displace first SPARC and then x86 in higher end servers - and that was a strategic threat Sun had to respond to.

The response was to replay AMD's victory, with the K5/6 line, over Intel during the 16/32bit transition. In that go-round, Intel's Pentium Pro incorporated leading edge 32bit ideas that should have won the market, but didn't because it wasn't fully backwards compatible with previous generation x86 gear. Thus when AMD introduced the K series combining 16 and 32 bit support, their chips ran old code much faster than Intel's did -and Intel was forced to put the Pentium Pro architecture aside to perpetuate 16bit instructions in the Pentium II and later CPUs.

In the current version of this, Sun engineers helped AMD put memory controllers directly into the silicon while extending the x86 instruction set to 64bits.

The result has pushed Intel into catch up mode again, forcing it - and therefore Microsoft - to adopt 64bit x86, slowed Itanium development to irrelevance, and incidently gave Sun a way to cash in on the x86/64 mini-boom while simualtaneously pushing Itanium further into its grave -by issuing hot new Opteron based products the other players have little choice but to try to match.

Meanwhile the market settled out, Java succeeded beyond anyone's wildest dreams, Sun produced Open Solaris, bought and released Open Office, developed the open identity management community and toolset, and re-emphasized the open nature of the SPARC specification by opening up its crown jewel - the T1 Ultrasparc architecture - in a public standard.

So where does that leave them today? Better positioned than anyone else to drive the next generation of hardware and software advancement.

And where's the hypocrisy and strategic confusion here? not in the executive suite at Sun, but you could try the editorial offices at the Economist