5th International ICST Conference on Collaborative Computing: Networking, Applications, Worksharing

Research Article

DDT: A distributed data structure for the support of P2P range query

Download457 downloads
  • @INPROCEEDINGS{10.4108/ICST.COLLABORATECOM2009.8356 ,
        author={Davide Carf\'{\i} and Massimo Coppola and Domenico Laforenza and Laura Ricci},
        title={DDT: A distributed data structure for the support of P2P range query},
        proceedings={5th International ICST Conference on Collaborative Computing: Networking, Applications, Worksharing},
        proceedings_a={COLLABORATECOM},
        year={2009},
        month={12},
        keywords={Aggregates Costs Data structures Distributed computing Peer to peer computing Proposals Publishing Telecommunication traffic Traffic control Tree data structures},
        doi={10.4108/ICST.COLLABORATECOM2009.8356 }
    }
    
  • Davide Carfì
    Massimo Coppola
    Domenico Laforenza
    Laura Ricci
    Year: 2009
    DDT: A distributed data structure for the support of P2P range query
    COLLABORATECOM
    ICST
    DOI: 10.4108/ICST.COLLABORATECOM2009.8356
Davide Carfì1,*, Massimo Coppola1,*, Domenico Laforenza1,*, Laura Ricci2,*
  • 1: ISTI, CNR, Via Moruzzi, Pisa, Italy
  • 2: Dipartimento di Informatica, Largo Bruno Pontecorvo, Pisa
*Contact email: davide.carfi@isti.cnr.it, massimo.coppola@isti.cnr.it, domenico.laforenza@isti.cnr.it, ricci@di.unipi.it

Abstract

This paper defines and evaluates a hierarchical distributed data structure, distributed digest trie, supporting range queries in P2P systems. Providing efficient support for these queries is currently a challenging research issue in the P2P field, as classical approaches based on distributed hash tables (DHT) are often not suitable for this kind of queries, due to the loss of locality introduced by the hashing function. Distributed digest trie exploits the DHT only to define a uniform assignment of logical identifiers to peers while each key is managed by the peer publishing it. Each peer is paired with the leaf of the trie corresponding to its logical identifier. An internal node of the trie stores a digest summarizing the keys published by the peers paired with the leaves of the tree rooted at that node. A proper mapping function is defined to map the internal nodes of the trie to the peers. The digests stored at the internal nodes are exploited to guide the search process for the resolution of the range query. Different aggregation techniques are proposed. A set of experimental results compare these techniques, evaluate the cost of dynamic updates of the data structure and the network traffic generated by the method.