The Dennis M. Ritchie Thesis Award 2020

The Dennis M. Ritchie Doctoral Dissertation Award was created in 2013 by ACM SIGOPS to recognize research in software systems and to encourage the creativity that Dennis Ritchie embodied, providing a reminder of Ritchie’s legacy and what a difference one person can make in the field of software systems research.

The Award committee for 2020 consists of Haibo Chen (SJTU), Nickolai Zeldovich (MIT), and Lidong Zhou (Microsoft Research).

The 2020 Dennis M. Ritchie Award winner is
Natacha Crooks

For her dissertation

A Client-Centric Approach to Transactional Datastores

at
University of Texas, Austin
advised by
Lorenzo Alvisi and Simon Peter

 

Natacha Crooks’ thesis achieves something rare: a new conceptual framework for client-centric consistency and two efficient systems built on those insights. The document for this attractive package is accessible (in part) to undergraduates and the advanced material is very clearly written. With the enduring popularity of consistency as a research topic in distributed systems for the past several decades it is surprising that a breakthrough as large as Natacha’s took as long as it did.

From the nomination letters

"[This dissertation] straddles principle and practice, in which design decisions have a rational, principled foundation."

"The dissertation makes contributions to the foundations of databases and distributed systems, and also contributes multiple system designs/implementations that demonstrate novel capabilities."

"This dissertation [addresses] a challenge that previously occupied, and sometimes stumped, Jim Gray, Phil Bernstein, and Barbara Liskov."

The 2020 Honorable Mention is
Anuj Kalia
Anuj Kalia

For his dissertation

Efficient Remote Procedure Calls for Datacenters

at
Carnegie Mellon University
advised by
David G. Andersen

 

It provides fundamentally grounded guidance about when and where we should split functionality between the CPUs and NICs in [the post-Moore's-law era]. The dissertation takes an approach that will age well, namely, that datacenter round-trips, measured in microseconds, will only grow increasingly more costly compared to performing processing locally within a machine. It then asks how to use the modern hardware options to efficiently get that processing to the CPUs nearest the data in ways that simultaneously meet the requirements for real deployments (congestion control, retransmission, etc.) and avoid both conventional and unexpected sources of overhead (OS networking stacks, peculiarities of the PCIe bus protocols, etc.).

From the nomination letters

"exemplifies the idea that going one level deeper to meticulously pick apart the fundamental source of performance bottlenecks allows systems designers and implementers make more informed decisions."

"shows that simpler, programmer-friendly designs can win out over approaches that drastically increase complexity and coupling"

"fundamentally advances the state-of-the-art in designing and building high-performance, low-latency distributed systems for modern center networks."

The awards were announced Tuesday, November 5th, in the ACM SIGOPS Award Ceremony at OSDI 2020.