Monday, December 02, 2019
1400 Wegmans Hall
Carnegie Mellon University
Pushing the Limits of Online and Offline Caching
Caches are pervasive in computer systems and often determine an applicationís end-to-end performance. Yet caching theory often makes unrealistic assumptions, forcing caching systems to rely upon best-effort heuristics that perform poorly and break easily. In this talk, I will describe recent work at CMU to better understand caching problems from a theoretical perspective and build the resulting insights into caching systems that improve performance or reduce cost. Many of the resulting techniques are applicable across the system stack, from processor caches of a few MBs to datacenter caches with hundreds of PBs. I will also discuss open problems that we have just started working on, or still have no idea how to solve.
Nathan Beckmann is an assistant professor in the Computer Science Department at Carnegie Mellon University, leading the CORGi research group. His PhD was advised by Daniel Sanchez at MIT and was recognized with the George M Sprowls award for an outstanding PhD thesis in computer science. He has received the NSF CAREER award and several Google Faculty Research Awards.