Kubernetes is one of the industry’s most important open source projects. It’s odd, therefore, that we’re not seeing much more participation from companies that have much to gain (and lose) from its rise.
True, a swelling population of enterprises are talking about contributing to Kubernetes, but when it comes to actually doing so? Not so much. Indeed, perusing the GitHub contribution data, you won’t find a single contributor from IBM, Pivotal (or VMware, its kissing cousin), Docker, or Oracle in the top 20 list. Unsurprisingly, you do find a heck of a lot of Google and Red Hat employees, as both companies have staked their futures on it, albeit in very different ways.
Red Hat, in particular, has dramatically upped its Kubernetes game of late, not only increasing its organic contributions to the project, but also paying a premium to buy CoreOS to gain expertise and product that will help it extend its Kubernetes lead in the enterprise. It’s the behavior of a company that understands just how high the stakes are for getting Kubernetes right.
Who made who?
So, who exactly are the top contributors to Kubernetes? While the list below doesn’t show the trend of individual contributor activity (you can get that by clicking here), it does provide a useful barometer for the top corporate contributors to Kubernetes:
- Clayton Coleman, Red Hat (1,155 commits)
- Brendan Burns, Microsoft (1,017 — but not much in the last year)
- Wojciech Tyczynski, Google (955)
- David Eads, Red Hat (864)
- Chao Xu, Google (831)
- Tim Hockin, Google (818)
- Stefan Schimanski, Red Hat (717)
- Daniel Smith, Google (697)
- Mike Danese, Google (666)
- Jordan Liggitt, Red Hat (590)
- Justin Santa Barbara, FathomDB (578)
- Marek Grabowski, Google (565)
- Derek Carr, Red Hat (556)
- Yu-Ju Hong, Google (454)
- Janet Kuo, Google (420)
- Nikhil Jindal, Google (384)
- Lantao Liu, Google (351)
- Jeff Grafton, Google (348)
- Zach Loafman, Google (331)
- Dawn Chen, Google (315)
There are at least two interesting takeaways from this list (and the commit trends shown in the link above). First, while Google employees remain active in Kubernetes development, Red Hat’s top contributors have been even more active over the past year (and, oddly, Microsoft’s Brendan Burns has largely been MIA as measured by commit activity). Rather than a sign of Google’s fading interest in the project, this is actually an indicator of effective community management by Google. A successful open source community can’t revolve around a single corporate sponsor.
SEE: Quick glossary: DevOps (Tech Pro Research)
Second, despite Red Hat recently ponying up $250 million to buy CoreOS to extend its Kubernetes expertise, no CoreOS developer makes the top 20 list. If we pull a different view of contributions, CoreOS shows up in the top 10 list of corporate contributors, which is pretty impressive, given its size. After all, CoreOS is a comparatively tiny startup. The fact that it (and FathomDB) show up at all is significant.
Red Hat definitely thinks so. At $250 million, Red Hat paid a lofty premium for a company whose revenue was negligible (with one former employee complaining of significant layoffs in the sales force as the company cast about for a viable go-to-market strategy). Beyond some revenue, what Red Hat really got was a team whose open source mindset is a near-perfect pattern match for Red Hat, not to mention a company that has been doing some of the more interesting tooling around Kubernetes.
All of which is worth much more than $250 million for Red Hat.
Justifying $250 million
How much more? Well, Red Hat currently generates more than $2 billion annually in revenues, but the key to its future is to add a few billion through things like Ansible (automation) and, in particular, Kubernetes (read: OpenShift). Red Hat executives have regularly touted the benefits of OpenShift to its business, leading to larger deals that pull along other products (like Red Hat Enterprise Linux).
SEE: Kubernetes: The smart person’s guide (TechRepublic)
Red Hat CEO Jim Whitehurst was clear on the company’s December 2017 earnings call: Going forward, it’s all Linux and Kubernetes for application development. Leading in both of these areas is critical to the company’s long-term success:
We have had two main selling points…. [O]ne is Kubernetes…[which] people see … as the … Linux equivalent and [the winner for] orchestrating containers. So, that’s obviously been helpful and as the second largest contributor to the Kubernetes project that’s positioned us well to be able to help drive roadmaps and confidently support our customers there.
The other is just lifecycle in the operating system. I mean, I think a lot of people forget that we have only security vulnerabilities. The user space of the operating system is inside the container. People like to … say, oh, the container is the source code for the application, but it’s also all of the user space dependencies. Over, 90% of the security vulnerabilities that are patched in Linux happen in user space. Therefore, if you are going to deploy an application that you plan to run in a production context, you need to make sure you have a vendor who you feel confident can support and patch the operating system components that are inside the container…. [O]ur credibility in … Linux [and] Kubernetes … [drive] our momentum in OpenShift.
And so $250 million becomes «cheap» compared to failure in containers. Red Hat historically was parsimonious in its acquisitions, having to shepherd its profitability. While that remains a concern for the company, it now generates enough profit that it can afford to invest heavily in areas that will pay big dividends down the road. Hence the relatively high valuations for InkTank, Ansible, and now CoreOS, with the last one showing the biggest gap between purchase price and the revenues the company was generating.
It’s the next RHEL for Red Hat, and that’s worth heavy organic engineering commitment to Kubernetes, as well as outsized acquisitions.