Sun Microsystems' recent purchase of Cobalt Networks tells me somebody at Sun has his or her head screwed on right. Sun is publicly positioning this move as a way to jump into the low-end server appliance market in a big way.
Cobalt is one of the best possible opportunities for Sun to do exactly that. In the long run, Cobalt helps Sun hedge its bets against the probability that its Solaris and Sparc-based systems will not hold the lead forever in the high-end Internet server space.
But Sun has three potential threats it must face in the future: Linux, Intel, and Windows 2000. Sun doesn't want to endorse Linux in a big way just yet, nor should it. Linux isn't ready to compete with Solaris on Sun's biggest boxes. And Sun can still get a lot of mileage out of its Sparc line of processors running Solaris.
Sun could take the attitude that Linux momentum will hit a brick wall when people discover it won't run well on big iron. But it can't count on that. A few weeks ago I said the Open Source Development Labs' (OSDL) plans to provide the needed resources and motivation to get Linux to scale efficiently on the best hardware. With the right person at the helm of OSDL, I predict Linux will start making its way onto high-end hardware within three years, tops - perhaps sooner.
Similarly, Intel-compatible hardware could catch up to the performance now offered by Sun hardware. (The operative word here is "compatible". Unless Intel gets its act together soon, AMD may be the leading Intel-compatible chip supplier by the time this plays out.) The endlessly delayed 64-bit Intel chips notwithstanding, the Intel platform has made nice strides recently in areas such as bus and memory speed.
At the risk of inspiring tons of flame mail, I believe Intel hardware still isn't even close to being competitive with Sun hardware. Assuming I'm wrong and Intel catches up quickly, Sun will have to lower its prices to remain an attractive alternative.
As far as Sun is concerned, who cares how this turns out now that Sun has Cobalt? If Intel can't match Sun's hardware performance, Sun will have built enough Linux credibility in a few years based on its Cobalt acquisition that it can start selling its next-generation Ultra Sparc servers running Linux instead of Solaris. Sun will keep its high margins and endorse the most popular platform.
If Intel-compatible hardware takes over, the Cobalt purchase gives Sun instant credibility in this potential future market, too. Put down that keyboard - I know that Cobalt's machines aren't all based on Intel-compatible chips. But Linux is the psychological link that will give Sun what it needs. Cobalt hardware runs Linux, and people tend to equate Linux with Intel. So any credibility that Sun builds with Linux automatically makes Sun a convincing supplier of Intel hardware if the company needs to go that route.
The best news of all for Sun is that no matter how the hardware picture develops, Sun's implicit endorsement of Linux by purchasing Cobalt puts yet another nail in the coffin of Windows 2000. This helps Sun eliminate the only threat about which it can do nothing.
It's currently fashionable to bash Sun Microsystems. But open-source advocates have much to be thankful for when it comes to Sun. Sun is going to open source Star Office 6.0. Sun gave the open-source Gnome desktop a mighty big publicity boost by adopting it for Star Office integration and for the Solaris desktop. But Sun's biggest contribution to open source could turn out to be purchasing Cobalt Networks.
* Nicholas Petreley is the founding editor of LinuxWorld (www.linuxworld.com). Reach him at firstname.lastname@example.org or e-mail comments for publication to david_beynon.com.au