Rashmi Vinayak - Assistant Professor, Computer Science Department, Carnegie Mellon University Communication systems
Date: - Location: Eurecom
Abstract: In large-scale data storage systems, disk failures are common. Hence, erasure codes are employed to store data in a redundant fashion to protect against data loss. In this setting, a set of k data blocks to be stored are encoded using an [n, k] code to generate n blocks that are then stored on distinct disks. The redundancy configurations (code
parameters) are typically set in a one-scheme-for-all-disks fashion. In this talk, I will
(1) present an analysis of failure traces from production clusters at Google, Backblaze, and NetApp showing that large-scale storage systems typically consist of devices with significantly varying failure rates, and
(2) show that substantial space-savings can be realized by tailoring code parameters to observed failure rates.
Traditional codes, however, suffer from high resource overheads in changing the code parameters on already encoded data. I will then
(3) present a theoretical framework to formalize the notion of "code conversion"---the process of converting data encoded using an [n, k] code into data encoded using a code with different parameters [n', k'], while maintaining desired decodability properties, and
(4) introduce "convertible codes", a new class of codes that enable resource-efficient conversion, bounds on their resource requirements, and explicit constructions.
Bio: Rashmi Vinayak is an assistant professor in the Computer Science department at Carnegie Mellon University. Rashmi received her Ph.D. from UC Berkeley in 2016, and was a postdoctoral scholar at UC Berkeley from 2016-17. Rashmi is a recipient of Sloan Research Fellowship 2023, Meta Research Award 2022, VMware Systems Research Award 2021, NSF CAREER Award 2020-25, Tata Institute of Fundamental Research Memorial Lecture Award 2020, Facebook Distributed Systems Research Award 2019, Google Faculty Research Award 2018, and Facebook Communications and Networking Research Award 2017. Her PhD thesis was awarded the UC Berkeley Eli Jury Dissertation Award 2016, and her work has received USENIX NSDI 2021 Community (Best Paper) Award, and IEEE Data Storage Best Paper and Best Student Paper Awards for the years 2011/2012. During her Ph.D. studies, Rashmi was a recipient of Facebook Fellowship 2012-13, the Microsoft Research PhD Fellowship 2013-15, and the Google Anita Borg Memorial Scholarship 2015-16. Her research interests broadly lie in computer/networked systems and information/coding theory, and the wide spectrum of intersection between the two areas.
