Improving Resource Efficiency in Cloud Computing

Cloud computing promises flexibility, high performance and low cost. However, despite its prevalence, most datacenters hosting cloud services still operate at very low utilization, posing serious sca lability concerns. This talk will discuss systems that improve datacenter utilization, while guaranteeing high performance for each submitted application.
A crucial system component to achieve this goal is the cluster manager; the system that orchestrates where applications are placed and how many resources they receive.

I will describe a new approach in cluster management that relies on two main insights; automating resource management by leveraging practical data mining techniques, and simplifying the responsibility of cloud users through a high-level declarative system interface that centers around performance constraints. I will describe Quasar, a practical cluster manager that leverages these techniques, and Tarcil a low-latency, sampling-based scheduler that enables millisecond-level decisions in clusters with tens of thousands of machines. We will also briefly talk about how these ideas translate to hybrid clouds, where private resources are combined with public systems. This work has had early adoption in industry, with production systems such as Twitter and AT&T introducing similar cluster management approaches.

Christina Delimitrou is an assistant professor at Cornell University working in computer architecture and cloud systems. Before that she was a post-doc researcher in the Computer Science Department at Stanford University, where she also graduated from with a PhD in Electrical Engineering in 2015. As part of her PhD work, she bu ilt practical systems for cluster management and scheduling in large-scale datacenters. Christina is the recipient of a Facebook Research Fellowship and a Stanford Graduate Fellowship. She has previously earned an MS from Stanford and a diploma in Electrical and Computer Engineering from the National Technical University of Athens.