Thanks to its free-speech and free-beer nature, the success of Linux is in the bag. Even Michael "How high did you say I should jump, Bill?" Dell has figured out that he can make more money by preloading a free operating system on his machines than by paying Microsoft a fee for every unit sold.
Those of you who still believe Microsoft and Windows 2000 are the safe choices on which to bet your company simply aren't paying attention. The LinuxWorld Conference & Expo packed the San Jose Convention Center to its limits this year, turning away dozens of vendors who were ready to pony up the cash for floor space. (To accommodate those in the rapidly growing Linux market, the expo will move to San Francisco's Moscone Center next year.) Dell, this year's keynote speaker, managed to turn many of his critics into fans with his convincing support for Linux. IBM is even including Linux as part of its high-volume enterprise-class Monterey AIX/Unix strategy.
Microsoft is an expert at momentum marketing, so we'll no doubt see Microsoft manufacture some numbers to try to convince people that Windows 2000 is catching on like wildfire. But Microsoft will have trouble making those numbers jibe with the way the vendor community is reacting to Windows 2000. The following says it all: IDG World Expo canceled this fall's Windows 2000 conference for lack of interest. If Windows 2000 is anything like wildfire it's because it is going down in flames.
The success of some of today's Linux companies is not so certain, however. Many new companies I visited on the LinuxWorld Expo floor have wonderful ideas and superb technology but have given little or no attention to a business plan. Even most of the established Linux vendors still haven't figured out how to make the kind of money expected of a public company.
That is precisely why I felt that the expo's highlight was the executive roundtable hosted by Patricia Sabga of the New Economy Watch on CNNfn. She grilled Larry Augustin of VA Linux Systems, Ransom Love of Caldera Systems, Art Tyde of Linuxcare, Volker Wiegand of SuSE, and Bob Young of Red Hat to find out where they will make money in the new market.
One distro for all?
Sabga's probing questions revealed that, most of all, her guests are ambivalent about standards. On the one hand, the distributors all recognize that independent software vendors, or ISVs, don't want to support multiple distributions. When Sabga pointed out that the Linux Standard Base (LSB) "isn't exactly progressing at the speed of light," the distributors insisted that wasn't an issue because Linux is already enough of a standard. But when she pressed them to demonstrate that Linux distributions aren't fragmented, they retreated into vague analogies about Volkswagens and sports cars. The problem with those analogies is that they cannot explain why a company like Oracle has to make separate deals with Red Hat and SuSE to ensure that its database will run well on either distribution. (One such deal between Oracle and SuSE was announced at this very LinuxWorld Expo.) And it is just plain dishonest of some of the distributors to say they all use the same Linux kernel. Distributions often have custom patches to the kernel, some of which are never folded into the mainstream kernel source. (There is nothing sinister about this practice: Distributors simply want to solve issues that have not been addressed in the mainstream kernel source. The patches may be folded into the mainstream kernel source, but that is up to the core kernel developers. Since the patches are open source, anyone is free to use them, so it is clear that the distributors are not trying to wrest control over Linux with proprietary extensions.) Interestingly, all the distributors said they supported a self-hosting standard -- that is, a complete standard base distribution that anyone could download and install -- to which they could add value. But minutes later they were hedging: such a self-hosting standard would probably never materialize, they said, and maybe it wasn't such a good idea anyway. It's obvious (although the distributors wouldn't admit it) that they found a self-hosting standard threatening because it would almost eliminate the barrier to entry for new Linux distributors. Anyone with a little start-up cash could become a distributor of the very version of Linux that major ISVs support.
The distributors should really have more confidence in their ability to add value to the basic distribution. They have the talent and resources to do so, but they are applying many of those resources to meet the demands of ISVs who don't want to wait for the kernel development community. As a result, many of the distributions are solving problems in ways that the kernel developers will probably override.
For example, the Oracle and SuSE deal assumes the use of the Reiserfs file system. But Reiserfs is not officially supported even in the new 2.4 kernel. If and when the kernel developers decide to add full default support for Reiserfs (if they haven't already and I missed it), they will probably do so in ways that differ from SuSE's. Wouldn't it make more sense for all the distributors to agree to provide the same type of Reiserfs support and then provide whatever resources are necessary to the kernel developers to make sure that the proper support is included in the next kernel? That way the kernel developers get more resources, the ISVs get what they want, and the distributors have to invest only a portion of their resources to get the same results.
Make no mistake: We will get a complete standard one way or another. At worst, some distributors will disappear completely or will acquire one another until the few left standing split up the market along server, workstation, or other lines using products based on the same basic distribution. But I'd rather see an independent standards body produce a complete self-hosting distribution. I believe that is the way more distributors will survive in the long run.
Is Linux enough?
Sabga also got a good debate going about revenues. It was clear that nobody is entirely sure that you can build a public company entirely on Linux support, although Art Tyde of Linuxcare was obviously optimistic since that's what Linuxcare is all about. I believe the jury is still out on that topic. But if I had to look into my crystal ball, I'd say that Linuxcare is going to have to develop expertise outside Linux to be most successful.
As I said at the start, the success of Linux itself is in the bag. But that doesn't mean alternatives cannot or should not also be successful. The BSD family is doing quite well and deserves even more attention than it gets. (I found it amusing that the default LinuxWorld Expo badges carried the BSD logo on the back thanks to some clever BSDi marketing.) AIX and Solaris aren't going away for a while. Even Windows NT is going to take some time to die out. Then there are all those open source and proprietary applications -- from Web servers to vertical applications -- that run on those OSs.
In other words, any organization that truly calls itself a support organization cannot be religious about either Linux or open source. It has to be religious about meeting the needs of its customers.
All in all, it was a fascinating LinuxWorld Expo in San Jose. I hope to see you at the next one in New York, and then back on the West Coast again in San Francisco.