Connecting Millions of Containers Spanning Dozens of Clusters - Service Discovery Challenges

Explore the challenges and solutions of connecting millions of containers across multiple Kubernetes clusters in this 35-minute conference talk by Laurent Bernaille and Antoine Tollenaere from Datadog. Dive into the complexities of service discovery in large-scale Kubernetes environments, particularly when workloads span multiple clusters and cloud providers. Learn about Datadog's infrastructure, consisting of millions of containers running across hundreds of clusters, and the strategies they employ for efficient service-to-service communication. Discover how Datadog addresses cross-cluster service discovery, targets deployments across multiple clusters, and implements solutions such as using Cilium to give pods native VPC IPs and running a global DNS architecture independent of clusters. Gain insights into emerging challenges and potential ideas for future improvements in managing large-scale, multi-cluster Kubernetes environments.