Traces in the Peer-to-Peer Trace Archive contains both dynamic and static data, on either peer level, swarm level, or community level of the measured P2P systems, and the dynamic and static data of is stored separately in each trace. For traces collected from swarm-based P2P systems like BitTorrent, data files in a trace have names like S6_peer_dynamic_data: S6 indicates this file contains data of swarm 5; peer indicates the file contains data depicts properties of the peer level of a P2P systems, and it can also be community or swarm; dynamic indicates the data depicts dynamic properties like peer events, and this field can also be static. For traces collected from non-swarm-based P2P systems like eDonkey and Gnutella, data files have names like chunk1_peer_dynamic_data: considering processing efficiency, traces are sometimes spitted into small chunks, and the chunk field indicates which chunk this data file is; and other fields have same the meaning with traces collected from swarm-based systems.
The data files in a trace are presented in a unified data format. Each line in data files contains ''timestamp'', ''swarm_id'', ''peer_id'', ''property_id'', ''iVal'', ''fVal'', ''sVal'':
For detailed descriptions of the data format, please refer to our publications:
B Zhang, A Iosop, J Pouwelse, P Garbacki. ''A Unified Format for Traces of Peer-to-Peer Systems''. Accepted by Workshop on Large-Scale System and Application Performance (LSAP). In conjunction with the International Symposium on High Performance Distributed Computing (HPDC 2009), June 10, 2009, Munich, Germany. [PDF] [BibTeX]
Boxun Zhang, Alexandru Iosup, and Dick Epema. "The Peer-to-Peer Trace Archive: Design and Comparative Trace Analysis". Technical Report PDS-2010-003, Delft University of Technology. [PDF] [BibTeX]
Side content