Sun makes a Linux U-turn
Cobalt developers were worried because of all the time they had put into developing applications for the version of Red Hat Linux that ran (and runs) on the Cobalt servers. Cobalt users were worried because the Cobalt developers were worried.
Sun's answer to the question indicated that it was indeed thinking about replacing Linux with Solaris, and the answer came from no less than Andy Ingram, senior vice president for Solaris. "Cobalt's (devices) are true appliances," Ingram said at the time. "You do not care about the operating system. If we were to convert these over to Solaris, the end-user wouldn't care. We don't have the energy to drive two operating systems."
Ingram also at the time stressed that since the Cobalt servers are true appliances, users do not care about the operating system. Of course he was right; on a true appliance, all you care about is the application. The operating system only becomes an issue when the application goes wrong.
As any Cobalt administrator will tell you, Cobalt servers are not quite true appliances. Sure, they look like appliances, they are marketed as appliances, and they smell like appliances. But they are not quite there; just go take a look at Cobalt's support site, where Raq owners are presented with a list of operating system updates and fixes, for instance adding simple password protection to directories. So most Raq administrators have become very familiar with some of the Linux configuration files on their 'appliance'. I do not recall the last time I had to update the software in my fridge.
The point of this is, that even though Cobalt devices are supposed to be appliances, users, as well as developers, do care about the operating system. And so when even Ed Zander, Sun's president, made it pretty clear that the Cobalt purchase was "not a Linux push," people got worried.
Well, it seems that both Zander and Ingram were wrong (I'm sure they were not misquoted) to suggest that Solaris would replace Linux on Cobalt, and everybody else was wrong to worry.
And the implications are significant because there are two pretty major u-turns here. First, rather than replace Linux with Solaris as was mooted 19 months ago, it looks as though Linux will actually be seeping out of the blue Cobalt boxes and into other Sun products. Second, Sun has from somewhere found the energy to support two operating systems after all.
And it's going to take a lot of energy. Key parts of Sun's plans echo two, quite separate major projects that have tied up massive resources in its competitors. First, its plan to deliver built-in Linux compatibility in key programming interfaces, commands and utilities, and user environments for Solaris are reminiscent of the IBM/Sequent/SCO Project Monterey, which fizzled quietly out to become AIX 5L. Just as IBM ditched SCO UnixWare in favour of Linux, so Sun too has seen the value in the massive numbers of Linux developers out there.
Second, Sun plans to extend its participation by offering key elements of Solaris. SGI anyone? The effort required to take bits of proprietary software and release them into the open-source community can be underestimated. Many would like to do it, but few can. Several years ago IBM/Lotus wanted to just this with SmartOffice. Just figuring out which bit of code was restricted by licenses upon licenses would have taken a million IP lawyers a million years. SGI tried it -- and indeed is still trying -- with key parts of the Irix operating system. It's the legal issues, not the technical issues, that have held up the work.
If one company can succeed in this, Sun can. But the question now has to be: will Sun ever replace Solaris with Linux?