Multi-layered Monitoring Framework with Indices Relationship Analysis for Distributed Service Performance Evaluation

System performance monitoring accumulates and analyzes performance information to detect and evaluate anomaly performance behavior. Recent advances in Internet based services have led to an increasing variety and complexity. Abnormal service operations adversely impact other services, making it extremely difficult to identify and resolve anomaly performance behavior. This work describes a novel multilayered framework to monitor multiple system layers simultaneously, including server layer, network layer, database layer, data processing layer, and user interface layer. By providing a comprehensive perspective of system monitoring, this work elucidates how performance indices across different system layers are related to information technicians attempting to explain abnormal system performances and resolve them. A state matrix structure is also developed for recording state information of system performance. Based on a state matrix database, the state information of new anomaly performance is inferred to resolve performance problems. A performance analysis and evaluation case study of the Social Media Analysis Platform demonstrates how to reduce administrative burden according to explain abnormal performance and resolve anomalies efficiently.