Cluster Reconciliation - Managing Resources Across Multiple Clusters

Explore the challenges and solutions for managing resources across multiple Kubernetes clusters in this conference talk. Learn about cluster reconciliation, including running services on multiple clusters, rolling updates, handling "snowflake" clusters, and gracefully adding or removing clusters from a group. Discover existing tools, techniques, and approaches used by upstream developers to address these challenges. Gain insights into the evolution of cluster management at scale, from hardcoded clusters to cluster groups and scheduling. Understand key concepts such as the Kubernetes object model, serverside apply, and various APIs including Cluster Inventory, Cluster Reconciler, Cluster Registry, and Workload API. Ideal for developers and administrators working with multi-cluster Kubernetes environments.

Intro
Why Multicluster
What kinds of Things
What Problems Are We Trying to Solve
What Does Failure Look Like?
The Kubernetes Object Model
Kubernetes Verbs
(Serverside) Apply
A Service in Kubernetes
Reconciliation Challenges
Evolution of Cluster Management with Scale
Hardcoded Clusters
Cluster Groups
Cluster Scheduling
The Cluster Inventory API
The Cluster Reconciler
The Cluster Registry
The Workload API
The Cluster Scheduler