2nd International IEEE Conference on Communication System Software and Middleware

Research Article

Performance Enhancement in Hierarchical Peer-to-Peer Systems

  • @INPROCEEDINGS{10.1109/COMSWA.2007.382555,
        author={Brajesh Kumar Shrivastava and Guruprasad Khataniar and Diganta Goswami},
        title={Performance Enhancement in Hierarchical Peer-to-Peer Systems},
        proceedings={2nd International IEEE Conference on Communication System Software and Middleware},
        publisher={IEEE},
        proceedings_a={COMSWARE},
        year={2007},
        month={7},
        keywords={Dynamic Host Configuration Protocol  Hierarchical DHT  Overlay Networks  P2P Networks  Round Trip Time},
        doi={10.1109/COMSWA.2007.382555}
    }
    
  • Brajesh Kumar Shrivastava
    Guruprasad Khataniar
    Diganta Goswami
    Year: 2007
    Performance Enhancement in Hierarchical Peer-to-Peer Systems
    COMSWARE
    IEEE
    DOI: 10.1109/COMSWA.2007.382555
Brajesh Kumar Shrivastava1,*, Guruprasad Khataniar1,*, Diganta Goswami1,*
  • 1: Department of Computer Science and Engineering Indian Institute of Technology Guwahati Guwahati - 781039, Assam, India
*Contact email: s.brajesh@iitg.ernet.in, gpkh@iitg.ernet.in, dgoswami@iitg.ernet.in

Abstract

Increased popularity of decentralized peer-to-peer (P2P) architecture comes with several challenges, like how to handle efficient lookup and query routing, frequent joining and leaving of nodes and scalability problem. Several key based deterministic routing approaches in a structured P2P networks have been proposed to overcome with these problems. Still, problems become worse if nodes are heterogenous in processing speed, power, stability and storage capability. To bridge the heterogeneity, we propose a 2-tier hierarchical distributed hash table (DHT) in which nodes are classified into three categories: temporary, stable and fully stable nodes. We have achieved intra-group lookup and query routing in O(log(|Gj|)) steps, where |Gj| is the number of (super)peers in destination group Gj, while inter-group lookup in O(1) steps. We have also saved at least 66.67% messages in repairing the routing table due to frequent joining and leaving of nodes in average case.