CNS Core: Small: Robust Performance Guarantee of Containerized Microservices in the Cloud

University Of Texas At San Antonio, San Antonio TX

Investigators

Abstract

Large-scale web services are increasingly being built with many small modular components (microservices), which can be deployed, updated and scaled seamlessly. These microservices are packaged to run in a lightweight isolated execution environment (containers) and deployed on compute resources rented from cloud providers. However, the complex interactions and the contention of shared hardware resources in cloud datacenters pose significant challenges in managing web service performance. This project will develop novel performance models and resource management solutions that can enable cloud platforms to provide robust performance guarantee for large-scale web services. This project will leverage multi-layered data collected from container-level resource usage metrics and virtual machine-level hardware performance counter metrics for accurate performance modeling. It will (1) develop probabilistic machine learning-based performance models that can quickly adapt to changing system dynamics and directly provide confidence bounds in the predictions even when the data is noisy and sparse. (2) It will directly incorporate predictive uncertainty obtained from the proposed models into the design and development of a robust resource scaling system and interference-aware container scheduling algorithm to manage the end-to-end tail latency of microservice request workflows in a resource efficient manner. The project will improve the performance, cost-efficiency, and scalability of a broad range of cloud-native applications. This can spur innovation in large-scale application development through increased adoption of cloud computing and cloud-native design principles. To this end, the project will develop open-source solutions for platforms such as Kubernetes. It will advance interdisciplinary education by developing a new course, which will contribute to a Cloud Computing Certificate jointly created by the University of Texas at San Antonio Colleges of Science, Engineering and Business. Data produced as a result of this project, including algorithms, models, software solutions, publications, and courseware, will be made publicly available at the project repository: http://www.cs.utsa.edu/~plama/microservices.html. The project data will be maintained and made available during the execution of the project and minimum three years after the project's ending date. Data will be stored on the host institution's dedicated web servers. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.