Editor's Note: Do the Retail Linux Numbers Mean Anything?
SuSE, Red Hat, and Mandrake lead the way
Establishing market share is always difficult in the Linux world. For starters, there's the issue of installations vs. sales: a single copy of Slackware Linux, for instance, can be installed on an unlimited number of servers and workstations. Secondly, there are a large number of "stealth" copies of Linux sold through bookstores and resellers like LinuxCentral; the bookshelves are creaking with Linux books with a CD or multiple CDs containing a full Linux distro. And polls of users done by internet.com show that corporate users by and large buy Linux directly from distributors, totally eschewing the reseller market.
That's why I sometimes don't take reseller data too seriously, but there are some conclusions that one can draw from the data. Take the results of the most recent retail market shares as researched by PC Data:
Red Hat 28.9%
Obviously, there are some caveats to be passed along with this. For starters, this covers retail sales of pure boxed copies of products: it doesn't cover book sales. (Macmillan puts out separate boxed editions Mandrake.) Secondly, without any pure sales figures, it's impossible to know how many units were sold -- which is a way to deduce revenues.
But I do believe that the sales figures do accurately represent -- to a degree -- how Linux is being used and by whom. SuSE is a strong player on the desktop, and because its packaging is so geared toward desktop users, I'm not surprised it gained nearly half of the retail sales. Because Red Hat is basically withdrawing from the desktop in favor of the server, I'm not surprised that it garnered less than a third of retail sales -- the surprise is that it did that well.
And Linux users are a pretty discrerning lot, if the sales of Caldera, Corel Linux, and TurboLinux are any indication. Caldera has announced that it will be focusing more on a server distribution, a developer version, and better integration with Open UNIX (i.e., UnixWare). Corel has already announced that it's shedding Corel Linux; couple that uncertainty with the fact that Corel Linux hasn't been updated for quite a while and you have a pretty good set of reasons to avoid Corel Linux. Finally, TurboLinux has withdrawn from the desktop world and is focusing on services and servers thanks to its merger with Linuxcare.
Still, I wouldn't read a whole lot into these retail numbers, as they basically only reflect who is going most strongly after retail sales. Server software simply isn't bought off of the shelf, and the server segment is where Linux is strongest.