Efficient processing of large graphs is challenging. With current data availability, real network traces are increasing in variety and volume, making it imperative to design solutions and systems based on parallel and distributed technologies. High-performance methodologies have the potential to benefit the graph processing community in much the same way they have advanced scientific computing. However, despite the success of high-performance computing in demanding scientific applications, graph analytics faces unique difficulties due to its data-driven nature, such as a high data-access to computing ratio and poor memory access locality.
Moreover, as the complexity of analysis increases, developing new solutions from scratch becomes increasingly demanding. This necessity underscores the importance of adopting software platforms—such as DBMS, frameworks, and libraries—to ease the development burden. Consequently, many data-oriented and graph-specific platforms have been proposed. We estimate that there are currently more than 180 platforms available for use in graph analytic tasks.