Which Metrics for Vertex-Cut Partitioning?

Hlib Mykhailenko (1) , Giovanni Neglia (1) , Fabrice Huet (2)

1 MAESTRO - Models for the performance analysis and the control of networks
2 SCALE - Safe Composition of Autonomous applications with Large-SCALE Execution environment

In this paper we focus on vertex-cut graph partitioning and we investigate how it is possible to evaluate the quality of a partition before running the computation. To this purpose we scrutinize a set of metrics proposed in literature. We carry experiments with the widely-used framework for graph processing Apache GraphX and we perform an accurate statistical analysis. Our preliminary experimental results show that communication metrics like vertex-cut and communication cost are effective predictors on most of the cases.

Domaines: Calcul parallèle, distribué et partagé [cs.DC]

Hlib Mykhailenko, Giovanni Neglia, Fabrice Huet. Which Metrics for Vertex-Cut Partitioning?. ICITST 2016 - The 11th International Conference for Internet Technology and Secured Transactions, Dec 2016, Barcelona, Spain. ⟨hal-01401309⟩