SERVICE MANAGEMENT USING DYNAMICALLY CALCULATED REQUESTS PER SECOND THRESHOLDS

原文（中国語）

Service management using dynamically calculated requests per second (RPS) thresholds include monitoring, over a span of time and across services instances of a service, a resource usage of a hardware resource and a requests per second (RPS) by the service instances obtain service instance data points that include resource usage metrics related to RPS numbers, and performing a regression of the service instance data points to obtain a current set of correlation functions between the resource usage and the RPS, and selecting a RPS threshold using the current set of correlation functions and a current amount of the hardware resource assigned to the service. Service management further includes continually monitoring the RPS for the service to detect when the RPS satisfies the RPS threshold and managing the service responsive to the RPS satisfying the RPS threshold.