Armada: Six months in the sandbox
By Jay Faulkner, Open Source Software Developer
Armada was accepted into the Cloud Native Computing Foundation (CNCF) sandbox in January 2023; the journey to get there, and what we’ve achieved since, has been no mean feat.
The good news is that Armada is well on the way to achieving its goal of becoming an industry standard for interoperability at scale. But getting a multi-Kubernetes-cluster batch scheduler off the ground, and ultimately where it is today, took a lot of work:
- First, it was not a simple software project. It required the collaboration and coordination of multiple stakeholders across a variety of organizations each with their own needs
- Second, the distributed nature of the project, combined with its sheer size and magnitude, made it a massive undertaking
We at G-Research wanted to share some of the work we’ve done, and are continuing to do, as part of building interest in Armada as it continues to integrate into the greater CNCF community.
Integrating with CNCF
The first steps in integrating with CNCF began before we were even admitted into the sandbox. Some of the steps were minor, such as creating an #armada channel in public CNCF Slack and ensuring that our developer documentation was fully up to date.
More daunting, however, was that Armada had to be moved from its original home, in the G-Research organization in GitHub, to a GitHub organization that only contained CNCF-related repositories. Many items integrate with, and are hardcoded to, a specific location of the code – finding and updating these is one example of the work required. It took about a week of work updating the automation, GitHub actions, and website as well as coordinating with the consumers of Armada at G-Research, before we were able to execute the move.
Getting listed in the CNCF Landscape made us a part of the community in name only. Then the real work began – building an open source team around Armada.
Building out our teams
There are two primary groups we targeted for building out that team: students who wanted an opportunity to learn via open source contribution and potential partners who might be interested in running Armada at scale and contributing to it.
Attracting junior-level coders who might be interested in learning more about open source contribution through Armada was an initial focus. There are a large number of grants and internship programs for open source projects to use for attracting developers.
We applied for, and were accepted as a Google Summer of Code (GSoC) project. The project, implementing a kubectl plugin for Armada, was highly sought after – one out of every five CNCF-related GSoC applicants applied to the Armada project.
Our team is hoping to see similar results with the Linux Foundation mentorship project to enable Armada to work with different database engines.
Building a developer community around Armada is only half the challenge. Without real organizations running Kubernetes in production and finding benefit from what we’ve built with Armada, we wouldn’t get a good sense of how it worked and what we needed to adjust, improve or make more usable as we moved forward.
For this reason, our other focus was on partners who might have interest in running Armada at scale. There’s a great place to find companies that run Kubernetes in production: Kubecon.
At Kubecon Europe in Amsterdam, there were many opportunities for us to share knowledge about Armada to interested parties. A lightning talk on Armada’s integration with Apache Airflow was attended by somewhere between 300 and 400 people and allowed a chance for people to learn about Armada. Many companies contacted us after this presentation interested in more information about how they could use Armada in their environments. We hope that over time, these potential partners will turn into collaborators with us on the Armada project.
Not only is Armada building a community, but it’s also joining one. Armada developers have become increasingly involved in CNCF ventures outside of Armada.
The product will continue to improve as long as people invest their effort into making Armada and its surrounding ecosystems better. Open source projects rely on the sustained efforts of others, always in collaboration, over time. Armada is no different.
In the immediate future, you can hear more about Armada at the OpenInfra Summit in Vancouver, where one of our developers will be presenting “Armada – Building a research platform on top of Openstack and Kubernetes.”
And you’re also welcome to join the Armada community – we’ll be happy to help get you started.