Recent hardware developments and application trends are challenging the long-standing server-centric data-center architecture. Two ideas have gained huge traction recently: resource disaggregation and serverless computing. Resource disaggregation breaks a computer server (either physically or virtually) into fine-grained, network-attached hardware resource units that can be shared by different applications. Serverless computing eschews "servers" by allowing users to directly deploy fine-grained programs ("serverless functions") that are triggered by external events.
With the natural synergy and importance of these two topics, the 4th Workshop on Resource Disaggregation and Serverless (WORDS 2023), to be held in conjunction with the 29th ACM Symposium on Operating Systems Principles (SOSP 2023), will bring together researchers and practitioners in operating systems, distributed systems, data-center networking, computer architecture, programming language, and domain-specific applications to engage in a lively discussion on a wide range of topics in the broad topicsv of resource disaggregation and serverless computing.
Keynote 1 (1:35pm-2:35pm)
Building The House On Strong Foundations: Abstractions for Efficient Resource Disaggregation and Serverless Computing
Malte Schwarzkopf (Brown University)
Disaggregation and serverless make an enticing promise: applications can access pools of resources without having to worry about machine boundaries or locality, and operators can fully utilize their expensive datacenter resources. However, the reality is quite far removed from this ideal. Serverless functions today couple resources, operators need to bin-pack them into fixed-size servers, and accessing remote memory often comes with a serious performance drop relative to using local memory.
I will try to separate the issues that hold disaggregation and serverless back into factors that are fundamental, and aspects that are under our control as system designers. For those aspects we can control, I will discuss new abstractions that help make resources more fungible, unstrand resources, and increase the efficiency of using remote resources. In particular, I will overview three abstractions that help build future applications on strong foundations: application-integrated far memory, soft memory, and decomposition of monolithic applications into granular "proclets". By building on these efficient low-level abstractions, I argue, we can achieve the goals of resource disaggregation and serverless in a way that leaves no performance on the table, but also provides building blocks for higher-level abstractions and libraries that help developers write applications that span servers and use datacenter resources wherever they are available.
Malte Schwarzkopf is an Assistant Professor of Computer Science at Brown University, where he leads the ETOS group. Malte's research is on new abstractions that deliver efficient, easy-to-use, and trustworthy computer systems. Recent projects include systems to make web services privacy-compliant by construction, high-performance remote memory, and soft memory, a new form of revocable memory. Malte is a recipient of the NSF CAREER award, a Google Research Award, Brown University's Henry Merritt Wriston Fellowship and Richard B. Salomon Award, and a class of 2023 Barrett Hazeltine Citation for Excellence in Teaching, Guidance, and Support. His past research received best paper awards at NSDI and EuroSys, as well as the EuroSys 2023 Test of Time Award. Prior to Brown, Malte was a postdoc with MIT's PDOS group and completed his PhD at the University of Cambridge. He is still getting used to living in a city not called Cambridge.
Session 1: Resource Disaggregation (2:35pm-3:20pm)
Shu-Ting Wang (UC San Diego); Weitao Wang (Rice University)
Anil Yelam, Stewart Grant, Enze Liu (UC San Diego); Radhika Niranjan Mysore, Marcos K. Aguilera (VMware Research); Amy Ousterhout, Alex C. Snoeren (UC San Diego)
Ruizhe Huang, Ding Li, Yao Guo, Xiangqun Chen (Key Lab of HCST (PKU), MOE; SCS, Peking University); Yuntao Liu, Yuxin Ren, Ning Jia, Xinwei Hu (Huawei Technologies)
Keynote 2 (3:45pm-4:45pm)
Microsecond-scale Datacenter Computing with RDMA: Characterization, Optimization and Outlooking
Haibo Chen (Shanghai Jiao Tong University)
Many online services are both latency-critical and throughput-oriented, which require datacenters to provision resources elastically and at microsecond-scale. Meanwhile, datacenters are now increasingly equipped with low-latency hardware like RDMA-capable SmartNICs as well as heterogenous on-chip and off-chip computing devices like CPU, GPU and NPU. In this talk, I will first present a characterization of emerging RDMA-capable NICs and SmartNICs and their interactions with computing devices, which uncovers many important performance features not systematically understood before. Based on this, I will summarize a set of optimization guidelines to efficiently harmonize RDMA with computing devices to provide microsecond-scale (tail) latency yet still high throughput. Then, I will present a set of case studies utilizing such guidelines to achieve significantly much better performance over state-of-the-art. Based on our experiences, I will outlook a set of open challenges and opportunities that may open new spaces for further evolution.
Haibo Chen is a Distinguished Professor of Shanghai Jiao Tong University, where he directs the Institute for Parallel and Distributed Systems (IPADS). His main research areas are operating systems and distributed systems. He received Best Paper Awards from ASPLOS, EuroSys and VEE, Test of Time Award from DSN, Best Paper Honorable Mention and Research Highlight Award from SIGMOD, Honorable Mention of The Dennis M. Ritchie Thesis Award (Advisor) from SIGOPS. He currently serves on the editorial board member and co-chair of Special Sections of Communications of the ACM, Program Committee of SOSP 2023/OSDI 2024, PC co-chair of EuroSys 2025, and the inaugural technical steering committee chair of OpenHarmony, an open-source operating system deployed on hundreds of millions of devices. He is an IEEE Fellow and an ACM Distinguished Member.
Session 2: Serverless Computing (4:45pm-6pm)
Dmitrii Ustiugov (NTU Singapore); Dohyun Park (UIUC); Lazar Cvetković (ETH Zurich); Mihajlo Djokic (IBM Research Europe); Hongyu Hè (ETH Zurich); Boris Grot (University of Edinburgh); Ana Klimovic (ETH Zurich)
Diogo Pacheco, João Pedro Barreto, Rodrigo Rodrigues (INESC-ID, Instituto Superior Técnico, Universidade de Lisboa)
Lazar Cvetković (ETH Zürich); Rodrigo Fonseca (Azure Systems Research); Ana Klimovic (ETH Zürich)
Rafael Alexandre, Rodrigo Bruno, Rodrigo Rodrigues, João Barreto (INESC-ID, Instituto Superior Técnico, Universidade de Lisboa)
Mafalda Sofia Ferreira, João Ferreira Loff, João Garcia (INESC-ID, Instituto Superior Técnico, Universidade de Lisboa)
Call For Papers
We solicit three types of papers: position papers that explore new challenges and design spaces, short papers that describe completed or early-stage work, and abstracts that summarize works published in the past two years.
Position and short paper submissions must be no longer than 5 pages including figures and tables, plus as many pages as needed for references. Abstracts of published works must be no longer than 2 pages, excluding references. Text should be formatted in two columns on 8.5x11-inch paper using 10-point Times-Roman font on 12-point (single-spaced) leading, 1-inch margins, and a 0.25-inch gutter (separation between the columns). New submissions must be double-blind. Abstracts of published works must be single-blind. Authors are allowed to post their papers on arXiv or other public forums.
Emmanuel Amaro - VMware Research
Rodrigo Bruno - Técnico (University of Lisbon)
Haibo Chen - SJTU
Christina Delimitrou - MIT
Kostis Kaffes - Columbia University
Aurojit Panda - NYU
Patrick Stuedi - Meta
Mona Vij - Intel