Characterizing global user download behavior on a large-scale satellite image distribution system

Data explosion in the 21st century has seen the growth of massive online content distribution systems. These incredibly large-scale systems utilize vast datasets to provide a myriad of utilities to users. In order to keep up with the high volume of complex, diverse requests being generated by their users, efficient management of data is of critical importance. However, understanding and categorizing the behavior of users worldwide is a complex task requiring a thorough understanding of the workflow they generate. In this paper, we examine and visualize over 5 million lines of real-world requests provided by the U.S. Geological Survey. We identify three distinct patterns in user requests that may be useful in developing intelligent prefetching methods.