A peer-to-peer architecture for multi-path data transfer optimization using local decisions

Data management (data retrieval and processing) performance in large-scale distributed systems (e.g. Grids, distributed databases, content delivery networks) is directly dependent on the efficiency and reliability of the communication architecture. The communication layer is responsible for transferring data between multiple source-destination pairs or for gathering the data from multiple sources to the processing nodes. In this paper we propose a peer-to-peer communication architecture for optimizing the efficiency, load balancing and reliability of the data transfers in the system, making local decisions only. We also present simulation-based experimental evaluation results.

[1]  Tom Kelly,et al.  Scalable TCP: improving performance in highspeed wide area networks , 2003, CCRV.

[2]  Antonios Argyriou,et al.  Distributed Optimization of Media Flows in Peer-to-Peer Overlay Networks , 2008, IEEE GLOBECOM 2008 - 2008 IEEE Global Telecommunications Conference.

[3]  B. Radic,et al.  Optimization of Data Transfer for Grid Using GridFTP , 2007 .

[4]  Elhadi M. Shakshuki,et al.  Agent-based peer-to-peer layered architecture for data transfer in wireless sensor networks , 2006, 2006 IEEE International Conference on Granular Computing.

[5]  Miguel Castro,et al.  SCRIBE: The Design of a Large-Scale Event Notification Infrastructure , 2001, Networked Group Communication.

[6]  M.I. Andreica,et al.  Towards a Communication Framework based on Balanced Message Flow Distribution , 2007, EUROCON 2007 - The International Conference on "Computer as a Tool".

[7]  Joachim Gudmundsson,et al.  Experimental study of geometric t-spanners , 2005, JEAL.