I-Corp: A Data-Driven Cloud Optimization Tool

George Washington University, Washington DC

Investigators

Abstract

The broader impact/commercial potential of this I-Corps project is the development of cloud management software using machine learning algorithms and optimization/reasoning to adapt to different application environments/demands, enable advanced forecasting and analytics, and automatically-generated, actionable recommendations. As more businesses and organizations continue to adopt a cloud-centric Information Technology (IT) infrastructure, there is an increasing demand for solutions to help optimize their cloud operations with the goal of improving the cost and operation efficiency of the infrastructure with advanced alerts, maximizing resource utilization and incident reporting/forecasting, and ensuring security compliance, e.g., access control and abnormal behavior detection. While there are some cloud management tools available on the market, they often lack the required intelligence to adapt to different application environments/demands, to enable advanced forecasting and analytics, or to automatically generate actionable recommendations on the fly. An automated and intelligent cloud management tool will address the needs of these customers. The tool will enable small- and medium-sized businesses to offer cloud-based services in a reliable and secure fashion. This I-Corps project will explore the translation of new developments of Reliability-as-a-Service (RaaS) technologies. RaaS are cloud management software technologies that address security and reliability. To enable fully automated, unsupervised operation of RaaS, the goal is to leverage multiple Recurrent Neural Networks (RNNs) to mine and decompose various trends (e.g., business growth and demand peaks) from rich cloud statistics. This mining and reduction process will allow the proposed artificial intelligence (AI) reasoning engine to identify complex underlying trends and analyze them in a synergistic fashion to obtain the most accurate analytics. The AI reasoning engine provides four functionalities in cloud management – Infrastructure Monitor, Cost Control, Efficiency Optimization, and Security Compliance. In addition, preliminary results show that the Infrastructure Monitor may predict cloud anomalies with an accuracy of over 91%. 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.