Over the last few months, IT editors and analysts have been competing to choose the 10 most significant technologies for business in the last 12 months. Most made some respectable choices - but naming Extensible Markup Language (XML) as the year's most significant business technology wasn't one of them.
I'm not arguing that XML hasn't had a big impact on business, although I would strongly suggest that it shouldn't have been rated Number One.
XML really isn't much more sophisticated than the label makers that are used to dial up letters and punch a name into a plastic strip. Yes, I know that when it comes to XML, it's not the label that matters but the standard. And I admit that, as a standard, XML has been one heck of a boon to business. But when I think of XML, I think of the Dewey Decimal System. And to me, Dewey just ain't a technology.
In fact, I'd bet people refer to XML as a technology these days only because Microsoft has succeeded in positioning XML as a replacement for Java. Microsoft undoubtedly hopes that by the time people notice "XML versus Java" doesn't make sense, Microsoft will have already shoved Java out of the picture and replaced it with the C# language. Good luck, Microsoft.
Java - now that's a technology. It's a language and it's a platform - two technologies in one! But Sun Microsystems has repeatedly fumbled its strategic handling of Java as a platform, and is doing so even now in the world of open source. Sun just doesn't know how to embrace the open source community, and it's hurting Java as Linux quickly penetrates the server market. Sun has to face the fact that it desperately needs Linux, both as the platform for Java servlets and as an underlying operating system for Java-enabled embedded devices.
I suspect the problem boils down to a fierce struggle within Sun, starting with the issue of Solaris versus Linux. How could a company so proud of its regal Solaris operating system possibly consider replacing it with Linux? I sympathise with the engineers in that respect - it would be difficult to dedicate your work to something, only to have it threatened by an upstart from the street.
The only major company with its own Unix that doesn't seem to have a problem adjusting to Linux is IBM. To its credit, IBM has fully embraced Linux at all levels.
I'm sure the engineers at Sun, SGI and HP would argue that, from a technical perspective, Solaris, Irix and HP-UX are superior to Linux and the BSD family. A few months ago they probably could have made that argument a convincing one; perhaps they still could if they picked apart the operating systems bit by bit. The real world is experiencing something quite different now. This item from a recent InfoWorld (PC World's sister publication in the US) news story should serve as a clear warning: "In December, Telia, Scandinavia's largest telecommunications company and ISP, installed a Linux-based IBM G6 mainframe to replace 35 Sun-based servers used largely to run its billing system."
And there's a lot more in that article that should send shivers down the spines of Sun, HP, SGI and others. Clearly business no longer perceives Linux as the toy operating system for PCs. According to InfoWorld, "IBM chairman Lou Gerstner announced in December that IBM will invest $US1 billion in Linux during 2001, a good portion of which will go toward the development, marketing and servicing of Linux on mainframes and supercomputer-class clusters."
So much for the myth that Linux can't scale like other systems. That myth - that Linux can't scale to 64 processors like Solaris - has been Sun's last refuge. It may still be true, but people are finding out that there's more than one way to scale - Linux is doing a fine job of scaling by running machines in parallel, as well as other methods.
It isn't all over for Sun yet. First, Sun needs to license Java under the GPL, even if it does so as part of a multiple licensing arrangement. Sun should also make a big public relations splash about donating engineering resources to harmonise Linux threading with Java threads. Sun could do that by contributing the necessary code to the Linux kernel source.
Sun has to decide quickly whether it wants to be a hero to the open source community, or a threat. So far, Sun has taken only a baby step with Open Office. That's wonderful, but it is not nearly enough. Source code for office productivity software is becoming extremely commonplace. What Linux really needs is a Java that runs quickly and flawlessly, and is open source in a way that the Linux developers embrace it. And Sun needs to overcome its obsession with Solaris so that it can compete with the likes of IBM when Linux starts to conquer the world.