Document nodelease feature (#10699)

content/en/docs/concepts/architecture/nodes.md

@@ -159,6 +159,20 @@ to be unreachable. (The default timeouts are 40s to start reporting
 ConditionUnknown and 5m after that to start evicting pods.) The node controller
 checks the state of each node every `--node-monitor-period` seconds.
 
+In versions of Kubernetes prior to 1.13, NodeStatus is the heartbeat from the
+node. Starting from Kubernetes 1.13, node lease feature is introduced as an
+alpha feature (feature gate `NodeLease`,
+[KEP-0009](https://github.com/kubernetes/community/blob/master/keps/sig-node/0009-node-heartbeat.md)).
+When node lease feature is enabled, each node has an associated `Lease` object in
+`kube-node-lease` namespace that is renewed by the node periodically, and both
+NodeStatus and node lease are treated as heartbeats from the node. Node leases
+are renewed frequently while NodeStatus is reported from node to master only
+when there is some change or enough time has passed (default is 1 minute, which
+is longer than the default timeout of 40 seconds for unreachable nodes). Since
+node lease is much more lightweight than NodeStatus, this feature makes node
+heartbeat significantly cheaper from both scalability and performance
+perspectives.
+
 In Kubernetes 1.4, we updated the logic of the node controller to better handle
 cases when a large number of nodes have problems with reaching the master
 (e.g. because the master has networking problem). Starting with 1.4, the node