Local Edge N+M high availability and scalability (no cloud connection required after setup)
1) Deploy Edges as local clusters such that they can suffer failure and there is a high availability strategy that doesn't require a cloud connection.
2) Edge scalability to horizontally add nodes to the Local Edge Cluster. Maybe it kind of behaves like Docker Swarm. i.e. Say I have a pretty intense preprocessor for data prep before handing it to ML Inference. That could get scaled out across multiple nodes in a Local Edge Cluster.
Our public preview of IoT Edge integration with Kubernetes is the first step towards running IoT Edge on clusters.