The release last week of TurboLinux's server clustering software raised concerns about whether the Linux community can prevent a divergence of versions - or forking - when vendors start making unilateral changes in the open operating system's kernel.
Linux's General Public License, which requires changes to Linux to be shared freely with the public, allows Linux to easily include unilateral innovations. But commercial versions of Unix have substantially diverged because their closed-source, proprietary licenses allow each vendor to compete on the basis of unique technological differences.
TurboLinux's software, TurboCluster, isn't the only Linux clustering effort, so there's no guarantee the Linux world will adopt its changes to the kernel. If competing software ends up in the mainstream kernel instead, it would be a fork, albeit a minor one, observers said.
TurboLinux's release is about six to eight months ahead of its competitors, according to analyst George Weiss at Gartner Group. But TurboLinux took the risk of forking, he said, when it decided to differentiate its distribution of Linux based on core technology, rather than higher-level applications or marketing. If it avoids that pitfall, he said, TurboLinux's gamble that users want Linux to scale to Unix-like heights of performance could succeed. "They have a shot at it," Weiss said.
TurboLinux's software was beta-tested by hundreds of users, including FDX in Memphis. FDX has acknowledged its role as a tester but hasn't commented on how the software performed or how it might use Linux clusters.
Joel Sloan, a systems administrator at Toyota Motor Sales USA, said TurboLinux has dim prospects. "I would hesitate to jump on board the TurboLinux bandwagon, since we know that [Red Hat Software] will have the same sort of clustering solution in a few months' time and the community at large will gravitate toward the more open, more mainstream solution.
"I predict that the TurboLinux solution will eventually be forgotten, unless their enhancements are folded back into the mainstream Linux code base. I would like to continue to see one mainstream Linux kernel," Sloan said.
According to analyst Tony Iams at DH Brown & Associates, Linux creator and kernel arbiter Linus Torvalds should take TurboLinux's changes seriously because high-availability clustering has been a major weakness of Linux.
There is precedent for Torvalds quickly deciding to incorporate changes to the kernel produced by commercial developers, Iams said.