Red Hat’s inverse-acquihire of CentOS makes sense

h-5c-a-1340
Red Hat and the CentOS project join forces, says the Red Hat official news release, “to accelerate community adoption and innovation for next-generation open source projects”. CentOS’s own announcement focuses on the new governance and long-term sustainability for the project with Red Hat sponsoring build processes and employing a number of core CentOS team members. The Red Hat plan for CentOS seems to be about having CentOS as the baseline platform for working with other communities on stuff like OpenStack, Gluster and OpenShift Origin. While CentOS, the project, remains independent with a firewall between it and Red Hat, like Fedora before it, it is being tied inexorably into the Red Hat ecosystem.

It seems that Red Hat has finally figured out that Fedora, despite its close ties and pioneering of Linux technologies, wasn’t creating a community around Red Hat Enterprise Linux. To use the tired car analogy, Fedora’s racing dune buggies, although testing out new technology for future models and great fun to drive, weren’t really bringing buyers to Red Hat’s range of business cars and vans.

The pace of new technology adoption outside the core of the operating system also meant that for Red Hat, developing say cloud software to work on Fedora would still mean an effort to backport it to RHEL or wait through the multi-year cycle for a major RHEL release. Now, with CentOS on board, that integration can be carried out on current RHEL-compatible CentOS releases vastly reducing the work needed to bring innovations into production on RHEL. And, if you recall, Red Hat has already shifted to a model of “Collections” allowing newer software to be shipped alongside their battle hardened enterprise Linux releases.

To play in open source, you need to be engaged with the community at all levels and Red Hat’s handling of RHEL has meant the community building work has gone to the groups who have taken the RHEL source and rebranded it, most prominently, CentOS. But Red Hat never officially condoned CentOS and a two arms length approach left the community distribution out in the cold. That cold zone also includes Oracle, who also take Red Hat’s source and create a RHEL compatible distribution of their own. And it is there, I suspect, that the real problem lay; for Red Hat to go on the offensive against Oracle would involve collateral damage to CentOS. With CentOS defensively under its wing, Red Hat may be able to bring more of a fight to Oracle’s ‘Unbreakable Linux”.

It’s going to be an interesting time for enterprise Linux with this move in place. It’s probably the smartest thing Red Hat could have done – inverse-acquihire the leaders of the RHEL cloning community. (Aquihiring is where you buy the company for the people, so I’m coining inverse-aquihiring for where you hire the people for the community). The competition will no longer be able to say “But there’s no ‘blessed’ free version of RHEL… our stuff is always free” while drafting up a services invoice. I expect the Fedora project to also get a freer hand in making more radical changes in future versions once CentOS starts taking on production-level integration.

But Red Hat and the CentOS team are going to have to remain smart – like all mergers, making the deal is just the start of “the time when things could go wrong”.