A read-only distributed hash table

Abstract

Distributed hash table (DHT) is an infrastructure to support resource discovery in large distributed system whereby data items are distributed across nodes in an overlay network based on a hash function. However, in commercial applications, data-item distributions may not be desirable due to the ownership issue and different economical interests among nodes. In this thesis, we present a DHT-based resource discovery scheme without distributing data items called R-DHT (read-only DHT). We further extend R-DHT to support multi-attribute queries with our Midas scheme (Multi-dimensional range queries). Our analysis shows that, firstly, R-DHT lookups are efficient and resilient to node failures. Secondly, the cost of query processing on R-DHT is low when the number of query results is small; this implies that the cost can be further reduced through selective data-item distributions. Overall, R-DHT is suitable for large distributed systems with strong commercial requirements in terms of data-item ownership and resource usage.