Web-Scale Range Queries in a Self-Organized Storage Service

Distributed solutions for the storage and retrieval of large amounts of data are necessary to handle the growing amounts of knowledge expected from future applications. The support for range queries provides much-needed expressivity, for example for queries on data annotated with location and time. We present a novel and scalable way for range query evaluation in our distributed storage system based on behaviour found in ants. Routing decisions in this system are taken on the basis of virtual pheromone paths leading from one node to another node, distinct for different data items. Range queries for single ranges can be evaluated by checking the pheromone paths for the entire range and forking the operation to several nodes if necessary. An aggregation method ensures the efficiency of the routing operations. We have evaluated our approach using a synthetic data set stored on 100 nodes, and executed various range queries while recording the amount of nodes involved along with the size of the result set.