This document discusses O(1) DHT designs that allow lookups in a distributed hash table to complete in one hop. It describes two approaches:
D1HT, which achieves this by having each node know all other nodes and maintain a full routing table, creating bandwidth issues during joins and leaves.
Kelips, which addresses this by having nodes periodically exchange partial routing tables to gradually build knowledge of the network, reducing bandwidth consumption but introducing some lookup latency.
Both aim to improve on the typical O(logN) lookup time in Kademlia-based DHTs by trading off bandwidth for speed, with potential applications in large-scale databases like Cassandra.