Scalable, Fast Cloud Computing with Execution Templates


Today, data analytics frameworks adopt one of two strategies to schedule their computations across workers. In the first strategy, systems such as Spark [3] use a centralized control plane, with a single node that dispatches small computations to worker nodes. Centralization allows a framework to quickly reschedule, respond to faults, and mitigate… (More)


Cite this paper

@article{Mashayekhi2016ScalableFC, title={Scalable, Fast Cloud Computing with Execution Templates}, author={Omid Mashayekhi and Hang Qu and Chinmayee Shah and Philip Levis}, journal={CoRR}, year={2016}, volume={abs/1606.01972} }