Model Based Control for Microservices Applications

摘要

Summary form only given, as follows. The complete presentation was not made available for publication as part of the conference proceedings. The move to cloud to leverage agility and scale initiated a fundamental shift from monolithic to microservices based applications. While they improved modular development, testing, and frequent improvements, microservices create significant challenges for Operations teams. These challenges result from application structure and behavior variability, shifting bottlenecks, and limited visibility and control of cloud infrastructure resources and services. Traditional performance management approaches such as queueing networks are not feasible for microservice applications given the lack of in-depth knowledge of the application structure and inability to handle the complexity of such systems at scale. Using heuristics that capture static relationships between performance and resources are also no longer applicable with highly dynamic virtual cloud resources. We propose a new model-based control for managing application performance. Relying on existing monitoring data and without instrumenting the application, we discover and build the application structure. Subsequently, using both machine learning (ML) and a priori knowledge of known services, we build a predictive application behavior model. The model is generated with sufficient granularity to detect anomalies that predict emerging performance problems. We then apply knowledge-based reasoning on insights from the anomaly analysis and dependencies within the application to recommend remedial actions to the infrastructure and services for problem resolution.