venue: NSDI
year: 2024
papers:
- title: 'MegaScale: Scaling Large Language Model Training to More Than 10, 000 GPUs'
  authors:
  - Ziheng Jiang
  - Haibin Lin
  - Yinmin Zhong
  - Qi Huang
  - Yangrui Chen
  - Zhi Zhang
  - Yanghua Peng
  - Xiang Li 0067
  - Cong Xie
  - Shibiao Nong
  - Yulu Jia
  - Sun He
  - Hongmin Chen
  - Zhihao Bai
  - Qi Hou
  - Shipeng Yan
  - Ding Zhou
  - Yiyao Sheng
  - Zhuo Jiang
  - Haohan Xu
  - Haoran Wei
  - Zhang Zhang 0003
  - Pengfei Nie
  - Leqi Zou
  - Sida Zhao
  - Liang Xiang
  - Zherui Liu
  - Zhe Li
  - Xiaoying Jia 0001
  - Jianxi Ye
  - Xin Jin 0008
  - Xin Liu 0086
  reason: "ByteDance's full production account of training LLMs at 10,000+ GPUs, with novel co-design of the network stack, fault tolerance, and collective communication to sustain near-linear scaling."

- title: 'Harmony: A Congestion-free Datacenter Architecture'
  authors:
  - Saksham Agarwal
  - Qizhe Cai
  - Rachit Agarwal 0001
  - David B. Shmoys
  - Amin Vahdat
  reason: "Proposes rethinking datacenter fabrics to eliminate congestion by construction rather than managing it reactively, achieving line-rate throughput without per-packet feedback from Cornell and Google."

- title: 'DINT: Fast In-Kernel Distributed Transactions with eBPF'
  authors:
  - Yang Zhou 0008
  - Xingyu Xiang
  - Matthew Kiley
  - Sowmya Dharanipragada
  - Minlan Yu
  reason: "Demonstrates that eBPF programs executing entirely inside the kernel can enforce linearizable, ACID-compliant distributed transactions at dramatically lower latency than user-space approaches."

- title: Making Kernel Bypass Practical for the Cloud with Junction
  authors:
  - Joshua Fried
  - Gohar Irfan Chaudhry
  - Enrique Saurez
  - Esha Choukse
  - Íñigo Goiri
  - Sameh Elnikety
  - Rodrigo Fonseca
  - Adam Belay
  reason: "Junction (MIT + Microsoft) is the first system to bring full kernel-bypass networking to multi-tenant cloud VMs without requiring application modifications or sacrificing isolation."

- title: 'SIEVE is Simpler than LRU: an Efficient Turn-Key Eviction Algorithm for Web Caches'
  authors:
  - Yazhuo Zhang
  - Juncheng Yang
  - Yao Yue
  - Ymir Vigfusson
  - K. V. Rashmi
  reason: "Surprising finding that a single-queue eviction policy requiring almost no state changes outperforms LRU and all modern approximations on production web-cache workloads."

- title: A large-scale deployment of DCTCP
  authors:
  - Abhishek Dhamija
  - Balasubramanian Madhavan
  - Hechao Li
  - Jie Meng
  - Shrikrishna Khare
  - Madhavi Rao
  - Lawrence Brakmo
  - Neil Spring
  - Prashanth Kannan
  - Srikanth Sundaresan
  - Soudeh Ghorbani
  reason: "A rare, candid production report from Meta on deploying DCTCP at hyperscale, surfacing unexpected interactions with heterogeneous hardware, cross-traffic, and operational constraints."

- title: 'Brain-on-Switch: Towards Advanced Intelligent Network Data Plane via NN-Driven Traffic Analysis at Line-Speed'
  authors:
  - Jinzhu Yan
  - Haotian Xu
  - Zhuotao Liu
  - Qi Li 0002
  - Ke Xu 0002
  - Mingwei Xu 0001
  - Jianping Wu
  reason: "Executes neural-network inference for traffic classification directly in the programmable switch data plane at line rate, eliminating the round-trip to a CPU-based classifier."

- title: 'Horus: Granular In-Network Task Scheduler for Cloud Datacenters'
  authors:
  - Parham Yassini
  - Khaled Diab 0001
  - Saeed Mahloujifar
  - Mohamed Hefeeda
  reason: "Offloads microsecond-granularity task scheduling decisions into programmable switches, reducing load-balancing latency by orders of magnitude compared to software schedulers."

- title: 'CAPA: An Architecture For Operating Cluster Networks With High Availability'
  authors:
  - Bingzhe Liu
  - Colin Scott
  - Mukarram Tariq
  - Andrew D. Ferguson
  - Phillipa Gill
  - Richard Alimi
  - Omid Alipourfard
  - Deepak Arulkannan
  - Virginia Beauregard
  - Patrick Conner
  - Philip Brighten Godfrey
  - Xander Lin
  - Joon Ong
  - Mayur Patel
  - Amr Sabaa
  - Arjun Singh
  - Alex Smirnov
  - Manish Verma
  - Prerepa V. Viswanadham
  - Amin Vahdat
  reason: "Google's production architecture for maintaining five-nines availability in Jupiter-scale cluster networks, detailing how control-plane redundancy and fast failover are achieved in practice."

- title: Revisiting Congestion Control for Lossless Ethernet
  authors:
  - Yiran Zhang
  - Qingkai Meng 0001
  - Chaolei Hu
  - Fengyuan Ren
  reason: "Identifies fundamental flaws in PFC-based lossless Ethernet that cause cascading head-of-line blocking and proposes a redesigned congestion control that avoids them."

- title: 'Sirius: Composing Network Function Chains into P4-Capable Edge Gateways'
  authors:
  - Jiaqi Gao
  - Jiamin Cao
  - Yifan Li
  - Mengqi Liu 0001
  - Ming Tang
  - Dennis Cai
  - Ennan Zhai
  reason: "Alibaba's production system for composing arbitrary NF chains into P4-programmed edge gateways, replacing a fleet of dedicated middleboxes and cutting per-packet processing cost significantly."

- title: 'CASSINI: Network-Aware Job Scheduling in Machine Learning Clusters'
  authors:
  - Sudarsanan Rajasekaran
  - Manya Ghobadi
  - Aditya Akella
  reason: "Shows that ignoring network topology when scheduling ML training jobs causes severe collective-communication contention, and that topology-aware co-scheduling yields substantial throughput gains."

- title: 'Cloudcast: High-Throughput, Cost-Aware Overlay Multicast in the Cloud'
  authors:
  - Sarah Wooders
  - Shu Liu
  - Paras Jain 0001
  - Xiangxi Mo
  - Joseph E. Gonzalez
  - Vincent Liu 0001
  - Ion Stoica
  reason: "Berkeley/Penn system that builds overlay multicast trees across cloud regions optimized for egress cost, enabling high-throughput data dissemination at a fraction of unicast cloud egress fees."