Etcd is a distributed data store that is used for example by Kubernetes to store the state of a running cluster. The data stores key-value pairs across multiple machines so that failures of a single machine are handled.
Etcd has been written for distributed systems like Kubernetes as a fault-tolerant and reliable data base. Clients can easily watch certain keys and get notified when their values change which allows scaling to a large number of clients that can reconfigure themselves when a value changes.
So far etcd has been an independent project and got now added to the CNCF as an incubating project ! Congratulations, etcd team!
At SUSE we include etcd as an essential part of our Kubernetes distribution SUSE CaaS Platform.