2017 IEEE International Conference on Cloud Computing Technology and Science (CloudCom) (2017)
Hong Kong, Hong Kong
Dec. 11, 2017 to Dec. 14, 2017
In this paper, we introduce a new methodology for automatic phase detection and characterization for applications running on the cloud. In contrast to existing approaches, our approach is novel in the fact that it is non-intrusive, more general (supports multiple programming languages), lightweight and can detect phase changes online as the application runs. We evaluate our approach for a number of C, C++ and Java application servers that are widely used in the cloud. Our method achieves a phase change detection accuracy upto 98.2% with an average detection delay of less than 0.01 seconds after the start or end of a phase. We also show a sample use case of our phase detection and characterization method for anomaly detection in the cloud.
Servers, Principal component analysis, Cloud computing, Hardware, Predictive models, Phase detection, Anomaly detection
A. Bhattacharyya, S. Sotiriadis and C. Amza, "Online Phase Detection and Characterization of Cloud Applications," 2017 IEEE International Conference on Cloud Computing Technology and Science (CloudCom), Hong Kong, Hong Kong, 2017, pp. 98-105.