Distributing and obtaining information on the network

[a-shine]

Distributing and obtaining information on the network can mostly be handled by a DHT protocol such as the Kadmlia but there doesn't seem to be much literature on how to deal with node failure. Node exit is fine e.g. if a node exists, it first transfers its information to another node and then shuts down but it fails it can't transfer its data. In this case, we need some form of data replication on the network that minimises data redundancy while still maintaining information in a stable way. It would be interesting to test with varying data replication count on an artificially unstable network what the probability of data retention would be.