A small-world DHT built on generalized network coordinates

Liz Ribe-Baumann, Kai-Uwe Sattler
2010 Proceedings of the 1st International Workshop on Data Semantics - DataSem '10  
Large-scale distributed hash tables (DHT) are typically implemented without respect to node location or characteristics, thus producing physically long routes and squandering network resources. Some systems have integrated round trip times through proximity-aware identifier selection (PIS), proximity-aware route selection (PRS), and proximity-aware neighbor selection (PNS). While PRS and PNS tend to optimize existing systems, PIS deterministically selects node identifiers based on physical node
more » ... location, leading to a loss of scalability and robustness. The trade off between the scalability and robustness gained from a DHT's randomness and the better allocation of network resources that comes with a location-aware, deterministically structured DHT make it difficult to design a system that is both robust and scalable and resource conserving. We present initial ideas for the construction of a small-world DHT which mitigates this trade off by retaining scalability and robustness while effectively integrating round trip times and additional node quality with the help of Vivaldi network coordinates.
doi:10.1145/1754239.1754286 dblp:conf/edbtw/Ribe-BaumannS10 fatcat:aaveomwzpzfe5mu72u7quscgxe