You are here

Using ML to Simplify IaaS Decision Making

Friday, 25. May 2018 - 11:00 to 12:30
Yehia Elkhatib (University of Lancaster, United Kingdom)
IRISA-INRIA salle Minquiers- campus de Beaulieu RENNES

Abstract : Users of cloud computing are overwhelmed with choice of providers and offered services. As such, many users select cloud services based on description alone. An alternative solution, behind which is a slowly swelling body of work, is based on decision support systems (DSS). In this talk, I will detail the motivation for such solutions, and I will explain how we built a data-driven DSS using machine learning methods to assist a cloud customer to arrive at optimal cloud deployment settings. Additionally, I will explain how we built an advanced DSS to significantly reduce the required amount of training data. This work is evaluated through intensive experimentation involving three real world applications over two major public cloud providers, namely Amazon and Google. We demonstrate how we are able to reduce required time and cost by 60%.

Bio : Dr Yehia Elkhatib is an Assistant Professor of distributed systems at the School of Computing and Communications of Lancaster University, UK, and a visiting professor at Ecole de Technologie Superieure, Montreal. He works to enable distributed applications to traverse infrastructural boundaries. In the context of cloud computing, this entails looking into interoperability and migration challenges, as well as related decision support issues. Yehia is the creator and chair of the international Cross-Cloud workshop series. Beyond the cloud, Yehia works on border-free network architectures in intent-driven systems, systems of systems, and information centric networks. He also works on advocating network-awareness which involves measuring networked systems, evaluating network protocols, and proposing new network management strategies.