Towards incremental updates in large-scale model indexes

Abstract

Hawk is a modular and scalable framework that supports monitoring and indexing large collections of models stored in diverse version control repositories. As such models are likely to evolve over time, responding to change in an efficient manner is of paramount importance. This paper presents the incremental update process in Hawk and discusses the efficiency challenges faced. The paper also reports on the evaluation of Hawk against an existing large-scale benchmark, focusing on the observed efficiency benefits in terms of update time; it compares the time taken when using this approach against the naive approach used beforehand, and discusses the benefits of combining the two, gaining improvements averaging a 70.7% decrease in execution time.