Network performance aware graph partitioning for large graph processing systems in the cloud

A wide variety of recent applications model their data in graphs/networks such as social networks, web graphs, and protein-protein interaction networks. Efficient processing for large graph data poses new challenges for almost all components of state-of-the-art data management systems. To list a few examples: (i) graph data are complex structures and cannot be efficiently stored as relational tables; (ii) the access patterns of large graph processing are complex, which results in inefficient disk accesses or network communications; and (iii) last but not least, to tackle scalability issues, graph processing must be efficiently distributed in a networked environment.