Pluribus Networks is now Arista Unified Cloud Fabric

Empowering Network Operations Teams with Pervasive Kubernetes Observability (26:14)

An increasing share of applications are built using containers that are orchestrated with Kubernetes. These cloud-native application environments tend to be highly distributed, with the majority of traffic comprised of East-West traffic flows within Kubernetes clusters. Network administrators typically have no visibility into these flows, increasing the difficulty of ensuring network and application performance.

This talk reviews innovative approaches to simplifying Kubernetes networking and enhancing visibility from application flows through the overlay to the physical underlay network. We review how switch-based, line-rate flow monitoring can provide full visibility into application flows within and between Kubernetes clusters, including session duration, data transfer and end to end latency. We then discuss mechanisms to extend this deep application visibility into servers and into public cloud environments for full visibility across a hybrid multi-cloud environment.