Scalability updates in Kubernetes 1.6: 5,000 node and 150,000 pod clusters

The community deserves a ton of credit for coming together or this investment. Special call out to CoreOS as the conversion from etcdv2 to etcdv3 which was was critical to the effort!

Disclosure: I work on Google on Kubernetes.

This is awesome. Congrats!

One thing that is missing (or hard to find) is the number of etcd nodes and the number of Kube master nodes that were used to back this cluster. Are there recommendations for this? For example, does adding more etcd or master nodes improve performance? My assumption is that scaling etcd or master nodes would allow the cluster to tolerate more failures and not improve performance, but I am not 100% certain. If there is an official doc on this, that would be great.