Experiment and Performance Analysis for Streaming Data on Multicore Platform.

Streaming data processing is one of the most important workloads that are handled by cloud gateways in supporting modern IoT applications to provide multimedia, browsing, management, monitoring and control functions. Diversified services place higher and higher requirements on the performance of the gateway computing platform, and the factors that affect the overall actual load and the performance of each part of the system are not direct. They depend on multiple dimensions and elements that restrict each other and need to be detailed according to the situation. In this work, the characteristics of streaming data workload on multi-core platform is analyzed based on benchmark experiments. The resource usage patterns of software and hardware is summarized and discussed in detailed through top-down performance factors. It helps to check pressure points, supports microbenchmarks, and creates performance models for further optimization.