A Decade of Kubernetes: How It Became the World’s Largest Open‑Source Project

Kubernetes Origins and Early Development

The first commit to the Kubernetes repository was pushed to GitHub on 6 June 2014. The initial snapshot contained 250 files and 47,501 lines of Go, Bash, and Markdown code. Early concepts stemmed from the need for higher‑level abstractions as hardware performance grew under Moore’s Law while application complexity increased.

Google engineers explored Linux‑kernel isolation mechanisms in the mid‑2000s. In 2006 Rohit Seth described containers as a structure for tracking and charging workloads such as memory and tasks. Google’s internal orchestration systems Borg and later Omega leveraged Linux containers. The rapid adoption of Docker (highlighted by Solomon Hykes at PyCon 2013) demonstrated the demand for an open‑source container orchestration platform, prompting a small team (Joe Beda, Brendan Burns, Craig McLuckie, Ville Aikas, Tim Hockin, Dawn Chen, Brian Grant, Daniel Smith) to start the Kubernetes project.

Key Milestones in the First Decade

Kubernetes 1.0 was released on 21 July 2015 and immediately donated to the Cloud Native Computing Foundation (CNCF). Early adoption was challenging, leading Kelsey Hightower to publish the “Kubernetes the Hard Way” guide in July 2016.

December 2016 – 1.5 added runtime plugins, early CRI support, Alpha Windows node support, OpenAPI, StatefulSets (Beta) and PodDisruptionBudgets (Beta).

April 2017 – Role‑Based Access Control (RBAC) introduced.

June 2017 – Third‑party resources (TPRs) replaced by CustomResourceDefinitions (CRDs) in 1.7.

December 2017 – Workload API reached GA in 1.9, stabilising Deployments and ReplicaSets.

December 2018 – Container Storage Interface (CSI) and kubeadm reached GA; CoreDNS became the default DNS server.

September 2019 – CRDs graduated to GA in 1.16.

August 2020 – Version‑support window extended to one year in 1.19.

December 2020 – Dockershim deprecated in 1.20.

April 2021 – Release cadence changed from four to three releases per year.

July 2021 – Widely used Beta APIs removed in 1.22.

May 2022 – Beta APIs disabled by default in 1.24; Dockershim removal caused upgrade confusion.

December 2022 – Batch and Job APIs redesigned in 1.26 to better serve AI/ML workloads.

Community contributions have grown to over 88 000 contributors, 15 000 code committers, more than 4 million commits, 158 000 issues and 311 000 pull requests.

Current Technical State (as of 1.31)

The upcoming 1.31 release will remove the built‑in cloud‑provider code, deleting roughly 1.5 million lines of code and reducing the core binary size by about 40 %. Vendor‑specific functionality is now provided via pluggable extensions such as CRDs or the Gateway API. The project is also migrating container image hosting to the community‑owned registry.k8s.io to lower bandwidth costs and improve performance. Users should update automation to pull images from the new registry.

Future Directions

Key focus areas include:

Enhancing support for heterogeneous hardware and large‑batch workloads that run on partitioned resources.

Continuing the removal of in‑tree vendor code in favour of extensible APIs.

Expanding the Serving Working Group (https://github.com/kubernetes/community/tree/master/wg-serving) to address AI/ML and high‑performance computing workloads.

New contributors are encouraged to follow the contributor guide at https://k8s.dev/contributors.

References

History blog: https://kubernetes.io/blog/2018/07/20/the-history-of-kubernetes-the-community-behind-it/

Kubernetes 1.0 lab tutorial: https://github.com/spurin/kubernetes-v1.0-lab

Registry migration blog: https://kubernetes.io/blog/2022/11/28/registry-k8s-io-faster-cheaper-ga/