See project at github acm wiki

• A DHT can be viewed as a dictionary service distributed over a network: it provides access to a common shared key->value data-store, distributed over participating nodes with great performance and scalability.
• From a user perspective, a DHT essentially provides a map interface, with two main operations: put(key, value) and get(key). Get will retrieve values stored at a certain key while put (often called announce) will store a value on the network. Note that many values can be stored under the same key.
• There are many algorithms to implement dht, where Kademlia DHT algorithm requires to contact only O(log(N)) nodes for a get operation, N being the number of nodes in the network. This property makes DHTs very scalable as demonstrated.

More info in Wiki:DHT

• Use go-lang to implement a DHT with basic functions
• Use this DHT to implement an easy application

• 70% - DHT tests
• 20% - Applications with stable network
• 10% - Bonus

• Learn Go

A Tour of Go
Go package docs
Books about Go

• DHT models

Chord
Pastry
Kademlia

• Related project framework

Dixie
CMU
MIT