Considering complex search techniques in DHTs under churn

Traditionally complex queries have been performed over unstructured P2P networks by means of flooding, which is inherently inefficient due to the large number of redundant messages generated. While Distributed Hash Tables (DHTs) can provide very efficient look-up operations, they traditionally do not provide any methods for complex queries. By exploiting the structure inherent in DHTs we can perform complex querying over structured P2P networks by means of efficiently broadcasting the search query. This allows every node in the network to process the query locally, and hence is as powerful and flexible as flooding in unstructured networks, but without the inefficiency of redundant messages. While there have been various approaches proposed for broadcasting search queries over DHTs, the focus has not been on validation under churn. Comparing blind search methods for DHTs though simulation we see that churn, in particular nodes leaving the network, has a large impact on query success rate. In this paper we present novel results comparing blind search over Chord and Pastry while under varying levels of churn. We further consider how different data replication strategies can be used to enhance the query success rate.