| Title | Predicting service metrics using real-time analytics |
|---|---|
| Speaker | Rolf Stadler (KTH Royal Institute of Technology, Sweden) |
| Abstract | Predicting the performance of cloud services is intrinsically hard. In this work, we pursue an approach based upon statistical learning, whereby the behavior of a system is learned from observations. Specifically, our testbed implementation collects device statistics from a server cluster and uses regression to accurately predict, in real-time, client-side service metrics for a video streaming service and a key-value store running on the cluster. The method is service-agnostic in the sense that it takes as input operating-systems statistics instead of service-level metrics. We show that feature set reduction improves prediction accuracy, while simultaneously reducing model computation time. We 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. This work is an ongoing collaboration with Ericsson Research. |
| Bio |
Rolf Stadler is a professor 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 is currently EiC of IEEE TNSM. Rolf’s research focus has been on integrated management of networks and systems. His group 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.
Click here to view Rolf Stadler's slides |
