Red Hat switched up CentOS to make it less of a Red Hat Enterprise Linux (RHEL) clone and more of a feeder project into RHEL (as Fedora was always supposed to be, yet wasn’t). Some people are mad, as Steven J. Vaughan-Nichols has written on sister site ZDNet. Some people, like former Disney employee Justin Garrison, think it sounds perfect (the hipper, slightly edgier version of RHEL). If you’re a billion-dollar company upset that Red Hat appears to be trying to charge for something you value, the founder of CentOS has a new way for you to get something for nothing: Rocky Linux.
But if you’re an open source entrepreneur wondering what this means for you, well, Chef cofounder and System Initiative CEO Adam Jacob has you covered. In a series of tweets, he walks through how Red Hat’s CentOS strategy can play out for you. (He should know, as the company he co-founded, Chef, last year open sourced everything.)
Open source all the things
Jacob’s first rule? Open it up. Completely. “If I do an open source strategy for a company ever again, I will own the upstream, it will be fully open source, and I’ll happily collaborate with anyone downstream.” But not just an open upstream–it’s also important to, “Produce a commercial distribution [and c]ollaborate on downstream non-commercial ones, in the open,” he argued.
SEE: Top 6 Linux server distributions for your data center (TechRepublic Premium)
What does he mean by “upstream” and “downstream”? In open source, think of the upstream as the parent, the head, the initial open source project. Downstream might be forks or distributions (packaging up of a particular build of the upstream code) of the upstream.
What Red Hat announced was basically that CentOS would move from being downstream to upstream. It becomes a place, as Jacob noted, that others like Facebook can collaborate with Red Hat in a way they simply couldn’t before (as Fedora wasn’t closely enough aligned with RHEL). CentOS as a downstream RHEL community was mostly one of users, of consumers, not of collaborators. It was somewhere to get RHEL, but rebranded CentOS, for free.
As such, Jacob pointed out, “They weren’t invested in it beyond using it.” And when someone removes the downstream they get mad “because it’s like someone threatened the water supply,” he argued. It’s therefore far better to condition people to participate as collaborators with an open source project, and through the commercial distribution to also condition users to become customers, if they want the certified distribution.
SEE: Clearing up the CentOS Stream confusion (TechRepublic)
Open source + cloud
One way that open source companies are doing this model to fantastic effect is by open sourcing their upstream and creating a cloud distribution (read: managed service). A variety of companies have embraced this model to greater or lesser extents.
Yugabyte, for example, ditched its Open Core model a year ago and open sourced 100% of its database code. A year later, CTO Karthik Ranganathan told me in an interview, “It increased our adoption like crazy,” growing the number of Yugabyte clusters 10x, but it also has dramatically accelerated their business without them losing any known pipeline. Could someone take that upstream and create a competitive downstream competitor? Of course. But no one should be able to out-Yugabyte on their home turf.
Or take Redis Labs. The company has fiddled with licensing over the last few years, but has kept core Redis completely open while encouraging a growing community (which includes downstream competitors) to lend a hand to improving the code. While Redis Labs doesn’t publish results, its business is booming, even as 10 or so other companies have created competitive downstream managed service offerings.
Which brings us back to Jacob: “Run an open upstream from the jump. Produce a commercial distribution. Collaborate on downstream non-commercial ones, in the open.”
That’s the strategy. That’s the magic. You don’t need to go Open Core or any other permutation of kind-of, sort-of open source. You can open source everything and just ensure you have a rock-solid managed cloud service. This reliance on cloud is what’s driving MongoDB, Confluent, DataStax, Redis Labs, and others to great success. It can be your model, too.
Disclosure: I work for AWS, but the views expressed herein are mine.