By Robert Cathey
At the most recent Tungsten Fabric developer summit in Seattle at KubeCon, community members gathered to map out new features and exchange use case best practices. Four of the event participants gathered over lunch, so we dropped a mic on the table and recorded their thoughts for the community to hear.
Valentine Sinitsyn of Yandex, Randy Bias of Juniper, Will Stevens of Cloudops, and Darien Hirotsu of Redapt shared their thoughts on:
- Tungsten Fabric vs Canal (Calico + Flannel): When to use which and why, and how they fit into the Kubernetes ecosystem; bridging the gap between legacy networks and cloud with Tungsten Fabric and Kubernetes.
- Tungsten Fabric and service meshes: L2 and L3 versus L7 for service meshes. Will they meet in the middle or overlap?
- Tungsten Fabric and Serverless: Can you bring up a port in 100ms? Tungsten can do serverless and bare metal and containers and more. Are there other ways to do this? What’s the positive impact of having refactored the Tungsten Fabric control plane in Kubernetes?
- Tungsten Fabric and DevOps/NetOps: Are all the options and choices in Tungsten Fabric good for DevOps or bad? What are the advantages of a consistent networking plane that’s abstracted away from the underlay?