Towards Complete Tracking of Provenance in Experimental Distributed Systems Research

摘要

Running experiments on modern systems like supercomputers, cloud infrastructures or P2P networks became very complex, both technically and methodologically. It is difficult to re-run an experiment or understand its results even with technical background on the technology and methods used. Storing the provenance of experimental data, i.e., storing information about how the results were produced, proved to be a powerful tool to address similar problems in computational natural sciences. In this paper, we (1) survey provenance collection in various domains of computer science, (2) introduce a new classification of provenance types, and (3) sketch a design of a provenance system inspired by this classification.