19-05-2010, 09:41 PM
Presented By:
Sandeep Ranade, and Mark S. Miller
Department of Computer Science
Abstract
Distributed hash tables have been used to store and manage data in a distributed environment. Popular approaches involve consistent hashing + caching. Hashing strategies try to randomly distribute data among nodes for reduction of hot spots. Hashbased routing tries to find short paths to the location where the data item is stored, or to a correct replica. Caching should make replicas of popular data closer to requesting nodes, and where the routing algorithm is likely to find it earlier. This speeds up lookups, and can also achieve some fault tolerance through redundancy of data. In Anthill, we try to exploit techniques used by ants for finding an efficient path to the food source, in a way that should be able to take into account network latency, access latency, processing overheads, prices, etc. Ants use pheromone trails and a form of positive feedback to reinforce certain