Research On Provenance Collection And Propagation In PASS Systems

Provenance-Aware Storage System(PASS)is a storage system that automatically collects and maintains provenance or lineage,the complete history or ancestry of an item.Provenance is traditionally the ownership history of an object.PASS automatically generates the DAG describing the relationship between processes running on it and the files stored in it.PASS tracks provenance at a file granularity.However,when the relation of object is complex,cycles will appear in DAG.PASS deal with this problem by detecting and breaking cycles.But the time complex of the algorithm is O(N~2),the work efficiency is much low,and then effect the efficiency of whole system badly.The paper introduces another algorithm,track provenance at a process granularity,to deal with the cycle problem.The time complex of the algorithm is O(N).PASS is at the beginning period of development,there is no idea of how to deal with transport problem.In other words,there is no any protocol to send provenance when PASS transports files.The paper introduces a preliminary method to deal with the transport problem,the design includes two steps:first step,after transporting the file to the client,PASS monitors the process,that transport files,to obtain the server's ip address and port information,and then PASS converts these information to the file's provenance and saves in the database.Second step,PASS gets the resource information of the file from the file's provenance and use the information to request server for the file's provenance.In the end,I make a series of experiment to compare the PASS and the modified PASS.According to the data of experiment,I find that modified PASS is faster then PASS by 50%.And in the aspect of provenance propagation,modified PASS can propagate provenance while transport files.