Aggregation Skip Graph: A Skip Graph Extension for Efficient Aggregation Query

Kota Abe, Toshiyuki Abe, Tatsuya Ueda, Hayato Ishibashi and Toshio Matsuura: Aggregation Skip Graph: A Skip Graph Extension for Efficient Aggregation Query, In Proceeding of the 2nd International Conference on Advances in P2P Systems (AP2PS 2010), pp.93–99, (Oct. 2010). (Best Paper Award)

Abstract

Skip graphs are a structured overlay network that allows range queries. In this article, we propose a skip graph extension called aggregation skip graphs, which efficiently execute aggregation queries over peer-to-peer network. An aggregation query is a query to compute an aggregate, such as MAX, MIN, SUM, or AVERAGE, of values on multiple nodes. While aggregation queries can be implemented over range queries of conventional skip graphs, it is not practical when the query range contains numerous nodes because it requires the number of messages in proportion to the number of nodes within the query range. In aggregation skip graphs, the number of messages is reduced to logarithmic order. Furthermore, computing MAX or MIN can be executed with fewer messages as the query range becomes wider. In aggregation skip graphs, aggregation queries are executed by using periodically collected partial aggregates for local ranges of each node. We have confirmed the efficiency of the aggregation skip graph by simulations.

Paper

Conference Site