Proxy Server Anomaly Detection Using API KPI Arbitration

The patent introduces an intermediary monitoring system that captures performance metrics through API calls rather than directly instrumenting the target servers. This intermediary layer (the monitoring platform) mediates between the need for performance data and the servers, obtaining information through standardized API interfaces that do not require modifying or directly interfacing with the server's core performance measurement mechanisms, thus avoiding additional load on the monitored servers.

The patent replaces traditional mechanical/direct performance monitoring methods (direct CPU, memory, cache measurements) with an information-based approach using API calls. Instead of directly accessing and measuring hardware resources, the system substitutes this with software-based information retrieval through APIs, which provides the necessary performance data without the overhead of direct hardware interaction.

The patent segments the anomaly detection function into multiple independent detection engines, each specializing in different aspects or types of anomaly detection. This segmentation allows each engine to focus on specific patterns or metrics, improving overall detection precision while maintaining manageable complexity through modular design. Each engine can be independently trained, deployed, and maintained.

The patent merges the outputs of multiple anomaly detection engines through an arbitration mechanism that combines their findings. This merging process integrates the strengths of different detection approaches, achieving higher overall precision. The arbitration engine synthesizes results from multiple sources, creating a unified anomaly detection output that leverages the collective capability of all engaged detection engines.