RIPPNET: Efficient range indexing in peer-to-peer networks

Write-heavy applications present a challenge to peer-to-peer indexing methods which need to update the index for each write operation. The costs incurred when the distributed index is updated becomes a bottleneck. Current distributed indexing methods are designed for indexing and retrieving single tuples, giving a very high update cost. In this paper we present a new approach to efficient peer-to-peer range indexing that employs indexing of ranges to reduce average update costs as well as providing efficient data localization and decoupling from data placement policies. Based on results from experiments, we demonstrate the applicability and significantly reduced update cost of the new approach.