Boris Grot

Serverless-native data analytics

Traditionally, large-scale data analytics jobs have run in a dedicated on-premise or cloud-based compute cluster. While effective in cases when the query load is steady, dedicated clusters can be inefficient from a cost and/or performance perspective if analytics jobs arrive sporadically or in sudden bursts. Serverless computing, with its extreme elasticity, rapid resource provisioning and usage-based billing, can fill the efficiency gap of cluster-based compute for irregular query arrival patterns. Alas, compared to a traditional cluster, serverless exposes a radically different compute substrate in the form of a vast pool of stateless workers that cannot directly communicate with each other. These differences motivate a need for a serverless-native analytics engine.

This talk will discuss our ongoing work toward that. I will describe the challenges and opportunities in serverless data analytics, and present our approach to navigating these with the Edinburgh Data Analytics Engine for Serverless (ENDLESS).

back to overview
Speaker Image
 

Biography

Boris Grot is a Professor in the School of Informatics at the University of Edinburgh, where he leads the EASE Lab. His research interests include server hardware and software stacks, networking, and datacenter-scale computing. Boris is a member of the MICRO Hall of Fame and a recipient of multiple awards for his research. Boris was the Program Chair for MICRO 2022 and the General Chair for HPCA 2024. "
Footer

Imprint / Privacy