/kubestellar

KubeStellar - a flexible solution for challenges associated with multi-cluster configuration management for edge, multi-cloud, and hybrid cloud

Primary LanguageGoApache License 2.0Apache-2.0





Multi-cluster Configuration Management for Edge, Multi-Cloud, and Hybrid Cloud


Generate and push docs    first-timers-only    Broken Links Crawler QuickStart test    docs-ecutable - example1    docs-ecutable - placement-translator    docs-ecutable - mailbox-controller    docs-ecutable - where-resolver    docs-ecutable - kubestellar-syncer    Join Slack

Overview

KubeStellar is an opensource project focused on concerns arising from multi-cluster configuration management for edge, multi-cloud, and hybrid cloud use cases:

  • Hierarchy, infrastructure & platform, roles & responsibilities, integration architecture, security issues
  • Runtime in[ter]dependence: An edge location may need to operate independently of the center and other edge locations​
  • Non-namespaced objects: need general support
  • Cardinality of destinations: A source object may propagate to many thousands of destinations.

Goals

  • Collaboratively design a component set similar to those found in the current kcp TMC implementation (dedicated Workspace type, scheduler, syncer-like mechanism, edge placement object definition, status collection strategy, etc.)
  • Specify a multi-phased proof-of-concept inclusive of component architecture, interfaces, and example workloads
  • Validate phases of proof-of-concept with kcp, Kube SIG-Multicluster, and CNCF community members interested in Edge

Areas of exploration

  • Desired placement expression​: Need a way for one center object to express large number of desired copies​
  • Scheduling/syncing interface​: Need something that scales to large number of destinations​
  • Rollout control​: Client needs programmatic control of rollout, possibly including domain-specific logic​
  • Customization: Need a way for one pattern in the center to express how to customize for all the desired destinations​
  • Status from many destinations​: Center clients may need a way to access status from individual edge copies
  • Status summarization​: Client needs a way to specify how statuses from edge copies are processed/reduced along the way from edge to center​.

Roadmap for the Project

We have defined and largely completed the PoC2023q1. The current activity is refining the definition of, and producing, the PoC2023q4. Goals not addressed in that PoC are to be explored later.

QuickStart

Checkout our QuickStart Guide

Contributing

We ❤️ our contributors! If you're interested in helping us out, please head over to our Contributing guide.

Getting in touch

There are several ways to communicate with us:

❤️ Contributors

Thanks go to these wonderful people:


Jun Duan

👀

Braulio Dumba

👀

Mike Spreitzer

👀

Paolo Dettori

👀

Andy Anderson

👀

Franco Stellari

👀

Ezra Silvera

👀

Bob Filepp

👀

Alexei Karve

👀

Maria Camila Ruiz Cardenas

👀

Aleksander Slominski

👀