Data-driven prediction of service metrics in real-time

Rolf Stadler. KTH – Royal Institute of Technology, Sweden

Abstract: We start with discussing the opportunities of using machine learning for network and cloud engineering and operation. We then focus on the specific problem of KPI estimation and present an approach that is based based upon statistical learning, whereby the behavior of a system is learned from observations. Following this approach, we collect device statistics from servers and switches on a testbed and use regression techniques to predict client-side service metrics for a video service and a key-value store. We further discuss design and implementation of a real-time analytics engine, which processes streams of device statistics and service metrics from testbed sensors and produces model predictions through online learning.

Bio: Rolf Stadler is a professor with the Department of Network and Systems Engineering at KTH Royal Institute of Technology in Stockholm, Sweden. He holds an M.Sc. degree in mathematics and a Ph.D. in computer science from the University of Zurich. Before joining KTH in 2001, he held positions at the IBM Zurich Research Laboratory, Columbia University, and ETH Zürich. Rolf Stadler is currently EiC of IEEE TNSM. His group has made contributions to real-time monitoring, resource management, and self-management for large-scale networks and clouds. His current interests include advanced monitoring techniques, as well as data-driven methods for network engineering and management.