RHEL clones: a surfeit of riches

RHEL is built with free software, and Red Hat fully understands both the conditions attached to free software licenses and the expectations of the development community. So the source for each RHEL release is promptly made available to the community. It is natural to expect that plenty of people would like to take advantage of the stability of and support behind RHEL without actually paying for a support contract; the availability of the source makes it possible to do exactly that. All that is needed is to rebuild those source packages, minus any Red Hat branding, and make the result available to the community.

Except that, in reality, the problem seems to be harder than that. Creating a proper build and distribution environment requires work and equipment. Quality control can be a lot of work, especially if strict binary compatibility with RHEL is desired. The security update stream must be followed constantly. And Red Hat does not always make the job of keeping up with their releases easy. So anybody wanting to make and support a proper RHEL clone must be prepared to invest a lot of time and infrastructure into the task.

Given that the desired end result of any RHEL rebuild effort - something that looks as much like RHEL as possible - is clear and unvarying, one would think that there would not be room for a large number of RHEL rebuild projects. There is not much space for ego or interesting new development, so it would make sense for everybody with an interest in this area to work together for the best end result. That is not how things have worked out, though. One need not look to far to find plenty of rebuilds out there:

• Oracle Linux is the most notorious of the commercial rebuilds. For the most part, Oracle has been working from the RHEL base with an eye toward binary compatibility. Over time, the company has started to add in some features of its own, including a newer kernel.

• CentOS is arguably the largest and best established of the free rebuild projects. Many commercial hosting providers offer CentOS installs, even if they have no other RHEL clones available. CentOS is clearly successful, but it is not the most community-oriented of projects. The project's leadership is jealous of its user base and seemingly unwilling to open things up to outsiders. At times the CentOS developers have failed spectacularly to keep up with Red Hat's releases, with the result that CentOS users have seen prolonged periods without security updates. That said, the rapid release of CentOS 6.2 suggests that the situation is improving.

• Scientific Linux has a significant and growing user base; it also has a small core of paid developers. This distribution has arguably done a better job of staying on top of security updates in recent times, but, as of this writing, it has not caught up to the Red Hat 6.2 release.

• Ascendos claims to be "is a free, community-supported, and professional-grade Linux platform." It is also vaporware, with no actual releases to date. Its development was initially led by Troy Dawson, who has also worked with Scientific Linux, but a job at Red Hat evidently put an end to that work. Troy's replacement, Douglas McClendon, stepped down in mid-December. Despite its claimed community orientation, Ascendos does not appear to have attracted many other developers, so the project now appears stalled.

• GoOSe, on its surface, looks a lot like Ascendos. It, too, claims to be "all about the community," but has not yet created much of an actual development community. The project had hoped to get a 6.0 alpha release out by the end of the year, but that did not quite happen.

  There have been discussions recently about cooperation between Ascendos and GoOSe, but nothing has been announced yet.

• ClearOS is a rebuild "built on source code from a prominent North American Linux vendor" offered by a group called "Clear Foundation." There is a lot of talk of community on Clear Foundation's web site, but "community" seems to consist mainly of access to a set of forums. By all appearances, ClearOS is meant to be a vehicle for a series of commercial support operations. The current ClearOS offering is based on RHEL 6.0, with a 6.2 release available as a beta.

• Fermi Linux is a rebuild of Scientific Linux. Its current 6.1 release adds a number of packages and tweaks some settings.

• PUIAS Linux is a rebuild published by Princeton University. It is currently based on RHEL 6.2 with a number of additional packages.

There are undoubtedly others, but, at this point, the picture should be clear: a lot of independent groups are putting a lot of effort into their own rebuilds of Red Hat Enterprise Linux. Some of them are more successful than others, but none are as good as they could be with a bit more focused effort.

Given the long list of existing distributions and the effort required to create and maintain a new one, anybody thinking of adding to the list should think long and hard about why that seems like a good idea. If the intent is to provide a Linux experience that is not possible with any of the existing distributions, perhaps there is an excuse for making a new one. But, in the case of RHEL rebuilds, there is simply no latitude for the creation of that new experience. The desired result is something that looks and acts like RHEL. Perhaps the various groups making their own RHEL clones would get better results if they were to build a single base distribution to work from. They could then compete fiercely to provide the slickest desktop theme, which is where all the interesting action is anyway.