This Active-Active application note focus on how to keep a piece of configuration fully synchronous across a cluster of ConfD’s while minimizing the availability sacrifice, i.e. locking the northbound interfaces when synchronizing the cluster, keeping the replicated configuration consistent at all times.
This can be done for a single leaf, a subtree, or the entire configuration (<-- unlikely use-case). Other parts of the configuration can be kept local or replicated through for example an “eventual consistency” approach for other parts of the configuration and operational data depending on the requirements.
See:
http://info.tail-f.com/confd_synchronous_clusters
Blog post:
http://www.tail-f.com/application-note-confd-active-active-fully-synchronous-ha-clusters/