For fault-tollerance data (key-ranges) are replicatad on R successive back-ends
FAWN-KV uses chain replication
updates sent to head of the chain (forwarded to others in chain)
queries sent to tail of the chain