Accelerating real-world streaming applications with Data Stream Processing frameworks

Abstract

The talk proposes the adoption of Data Stream Processing frameworks (Storm, Flink, WindFlow, …) to perform real-time processing in Cyber-Physical Social Systems (e.g. Complex Event Processing). The idea is to improve performance (bandwidth and latency) by exploiting edge resources to parallelize the computation. Proper heuristics are required to find the optimal allocation of the operators of a streaming application onto the graph of edge nodes.