Graph Analytics: Complexity, Scalability, and Architectures

P. Kogge
DOI: https://doi.org/10.1109/IPDPSW.2017.176
2017-05-01
Abstract:Big Data as expressed as "Big Graphs" are growing in importance. Looking forward, there is also increasing interest in streaming versions of the associated analytics. This paper develops an initial template for the relationship between "traditional" batch graph problems, and streaming forms. Variations of streaming problems are discussed, along with their relationship to existing benchmarks. Also included is a discussion of classes of parallel architectures (including newly emerging ones) and how such kernels are liable to scale on them. Preliminary projections for some of these systems is presented.
What problem does this paper attempt to address?