In 2014 Google packaged their experience building massive and automated container clusters into a remarkable system of open source software known by a Greek word for “helmsman”: Kubernetes. Successor to the internal Borg system and its research-oriented successor, Omega, the introduction of Kubernetes garnered massive interest. The open source project quickly took on new developers outside Google.
Kubernetes has since become a diverse and broad community. At the Google Next conference last week, it was reported that there have been over 43,000 commits in the last three years, which is estimated in amounting to over 400 years worth of work. The rapid expansion of Kubernetes is due largely to this substantial increase in contributors. In 2014, the project started with just 150 active monthly contributors. Today, this has risen to over 1,500 unique all-time contributors spanning every time zone.
At CoreOS, we are proud of our role in helping build the Kubernetes community. We chose to be part of the Kubernetes community because it is the right technology to enable our mission to secure the internet and deliver self-driving infrastructure.
Working in the community has also allowed us to be more in tune with customer needs. We are better prepared to support customers and to transform enterprise requirements into usable solutions. We’re proud to be one of the more influential contributors to the Kubernetes community. Measuring this influence within an open source project is tricky as it is more than a simple count of pull requests. We view it as a measurement of time spent architecting, contributing, and evangelizing the entire project.
This is achieved by introducing others to Kubernetes via workshops, participating in SIG meetings, and more. This post is a reflection on the ways we have participated and influenced this awesome project over the years.
Over the last three years, CoreOS has helped to lead the technical direction of the project in several ways. We have pioneered several new concepts in Kubernetes, including: self-hosted, using Kubernetes to manage Kubernetes; self-driving, automated updates and controls; and operators, which represent operational knowledge in software.
CoreOS devs help guide these architectural decisions by remaining very active in the Kubernetes Special Interest Groups, where we lead or co-lead six and participate in 19 of the 25 groups. We also have filled many formal leadership roles in the community. For instance, the release manager for the upcoming 1.6 release will be Dan Gillespie, a CoreOS engineer. This is the first time a resource outside of Google has filled this critical role.
Back in 2014 our CTO, Brandon Philips, acted as one of the first product managers outside of Google. Brandon is a recognized tech leader and by realizing the promise of this important project, he lent K8s exposure and credibility. Both are important for a young project.
Brandon has continued his heavy involvement, and still contributes to the direction and governance of K8s. Kubernetes became an independent entity within the Cloud Native Computing Foundation (CNCF), due in part to his leadership. Brandon also played an important role in defining the core project features, and helped to create the Kubernetes Incubator program for non-core projects. He remains a key advisor to the governance group and is driving standards as a chair of the technical oversight committee of the Open Container Initiative (OCI) and active member of the CNCF.
From its first release in 2014, we have been committing to Kubernetes. The first public release of Kubernetes was version 0.2. We were heartened to see that even the likes of Google are subject to off-by-one (or 0.1) errors. Kidding aside, in that early release CoreOS engineers made two commits. We’ve been contributing at a high velocity ever since, dedicating the majority of our engineering resources to Kubernetes related projects. As a result, we’ve become the largest non-public corporate contributor (as measured by pull requests).
The contributions don’t stop at PRs. We’ve developed critical components like etcd, which acts as the primary datastore of Kubernetes; storing and replicating all Kubernetes cluster state. When Redspread joined CoreOS, much of their work was similarly donated. Some of that code eventually merged and evolved into Minikube.
We believe that with leadership also comes with the responsibility to teach and empower others. We pursued this by providing Kubernetes training to over 1,000 people in 2016. To date we have organized over a dozen of container related meetups and speak all over the world at conferences like FOSDEM and KubeCon.
To punctuate that last note about conferences, you may have noticed that folks from CoreOS appear on stage fairly regularly. Brandon is easily recognizable, in part because of how regularly he takes the stage. In the latter parts of 2016, he attended KubeCon and delivered three keynotes. He announced self-driving Kubernetes, improvements to etcd and a hopeful look forward towards federation. He’s since attended many more events, spreading the Kubernetes gospel throughout the land.
Other notable CoreOS faces may include Brian “Redbeard”, who recently spoke at ShmooCon. He also joined Google Cloud’s panel during RSA conf. You’re most likely to recognize him as the MC of Tectonic Summit and CoreOS Fest.
Tectonic Summit and CoreOS Fest are two events we run, which also help educate and get the word out about Kubernetes. Tectonic Summit is targeted at enterprise users. Conversely, CoreOS Fest is the Yin to Summit’s Yang; so, its focus is on community.
Showing the value of Kubernetes is critical. As more and more organizations and individuals adopt it as their platform for orchestrating containers, Kubernetes gains momentum and further community involvement. This snowball effect is critical for its long term viability and it is moving quickly.
We’ve been able to bolster the Kubernetes community with our broad range of experience, but Kubernetes has given a lot back to CoreOS as well.
The work we do with container technology spans the ecosystem. CoreOS Container Linux, Quay, rkt and Clair are all container centered projects. Our focus on distributed systems infrastructure and the expertise that we have accrued allows us to make meaningful contributions to the space.
This trust from the community has translated into trust from major cloud vendors. An example of this is Amazon’s endorsement of CoreOS’s Tectonic platform as the recommended way to install and manage Kubernetes on AWS.
This strong and positive relationship with vendors has led many Global 2,000 companies relying on CoreOS technologies to build infrastructure that operates at hyperscale. Major companies using open source projects developed by CoreOS include: Alibaba, Amazon, Apple, Baidu, Google, Microsoft, Salesforce, Tencent, Ticketmaster, and Yandex.
From the very beginning of the project, CoreOS has worked to make Kubernetes the leading container orchestration platform that it is today. Through our technical contributions as well as our involvement in the governance and direction of the project, we have sought to provide the most technically and architecturally sound open source solution to the community and enterprise. We are excited about the direction of the project, and look forward to continuing to help it expand and mature in coming releases.
CoreOS is hiring more team members to join us on the mission to develop self-driving container software in the upstream Kubernetes community. Join CoreOS on the journey with Kubernetes and learn about joining the team at coreos.com/careers.
If you are at KubeCon Berlin on March 29-30, learn more with us at the CoreOS booth and various sessions presented by the team working on Kubernetes.
If you are in San Francisco, join us at an SF meetup on Thursday, March 30 at Operator Hack night.