NSDI ’21
as the organisers put it
The 18th USENIX Symposium on Networked Systems Design and Implementation (NSDI '21) will take place as a virtual event on April 12–14, 2021. NSDI focuses on the design principles, implementation, and practical evaluation of networked and distributed systems. Our goal is to bring together researchers from across the networking and systems community to foster a broad approach to addressing overlapping research challenges.
| title | author | length |
|---|---|---|
| Accessing Cloud with Disaggregated Software-Defined Router | Hua Shao, Xiaoliang Wang, Yuanwei Lu, Yanbo Yu, Shengli Zheng, Youjian Zhao | 12m |
Tencent Cloud replaces monolithic routers with a Disaggregated Software-defined Router that splits access (commodity switches), fast software forwarding… | ||
| CodedBulk: Inter-Datacenter Bulk Transfers using Network Coding | Shih-Hao Tseng, Saksham Agarwal, Rachit Agarwal, Hitesh Ballani, Ao Tang | 12m |
Claims a 1.2–2.5× throughput boost for inter-datacenter bulk transfers by converting multicasts into algebraic mixes using linear network coding and running as… | ||
| Twenty Years After: Hierarchical Core-Stateless Fair Queueing | Zhuolong Yu, Jingfeng Wu, Vladimir Braverman, Ion Stoica, Xin Jin | 12m |
Implements Core-Stateless Fair Queueing on programmable switches and generalizes it to Hierarchical CSFQ that approximates hierarchical fair queuing while… | ||
| Breaking the Transience-Equilibrium Nexus: A New Approach to Datacenter Packet Transport | Shiyu Liu, Ahmad Ghalayini, Mohammad Alizadeh, Balaji Prabhakar, Mendel Rosenblum, Anirudh Sivaraman | 12m |
NSDI '21 - Breaking the Transience-Equilibrium Nexus: A New Approach to Datacenter Packet Transport
Shiyu Liu and Ahmad Ghalayini, Stanford University… | ||
| Running BGP in Data Centers at Scale | Anubhavnidhi Abhashkumar, Kausik Subramanian, Alexey Andreyev, Hyojeong Kim, Nanda Kishore Salem, Jingyi Yang, Petr Lapukhov, Aditya Akella, Hongyi Zeng | 11m |
Facebook describes a BGP-based data-center routing design that scales across hundreds of pods by reusing ASN blocks, hierarchical route summarization to cut… | ||
| Orion: Google's Software-Defined Networking Control Plane | Andrew D. Ferguson, Steve Gribble, Chi-Yao Hong, Charles Killian, Waqar Mohsin, Henrik Muehe, Joon Ong, Leon Poutievski, Arjun Singh, Lorenzo Vicisano, Richard Alimi, Shawn Shuoshuo Chen, Mike Conley, Subhasree Mandal, Karthik Nagaraj, Kondapa Naidu Bollineni, Amr Sabaa, Shidong Zhang, Min Zhu, Amin Vahdat | 12m |
Orion runs Google’s second‑generation SDN control plane across Jupiter and B4, built as microservices around a pub-sub Network Information Base (NIB) and… | ||
| Metha: Network Verifiers Need To Be Correct Too! | Rüdiger Birkner, Tobias Brodmann, Petar Tsankov, Laurent Vanbever, Martin Vechev | 12m |
Metha automatically tests network analysis tools by generating syntactically and semantically valid configurations in a two‑stage generator, running each… | ||
| Finding Invariants of Distributed Systems: It's a Small (Enough) World After All | Travis Hance, Marijn Heule, Ruben Martins, Bryan Parno | 11m |
Swiss automates construction of inductive invariants by exhaustively searching a bounded candidate space, using counterexample-guided synthesis and SMT checks… | ||
| Avenir: Managing Data Plane Diversity with Control Plane Synthesis | Eric Hayden Campbell, William T. Hallahan, Priya Srikumar, Carmelo Cascone, Jed Liu, Vignesh Ramamurthy, Hossein Hojjat, Ruzica Piskac, Robert Soulé, Nate Foster | 12m |
Avenir automatically synthesizes control-plane operations to make diverse data planes behave uniformly, proving soundness and completeness and implementing the… | ||
| Don't Yank My Chain: Auditable NF Service Chaining | Guyue Liu, Hugo Sadok, Anne Kohlbrenner, Bryan Parno, Vyas Sekar, Justine Sherry | 13m |
AuditBox is a deployed system that uses trusted execution environments (Intel SGX enclaves) plus a per-NF verification shim to produce continuous, provable… | ||
| Contracting Wide-area Network Topologies to Solve Flow Problems Quickly | Firas Abuzaid, Srikanth Kandula, Behnaz Arzani, Ishai Menache, Matei Zaharia, Peter Bailis | 13m |
NCFlow contracts the WAN into clusters and replaces the full multi-commodity LP with a tiny aggregated inter-cluster flow plus parallel per-cluster… | ||
| Cost-effective Cloud Edge Traffic Engineering with Cascara | Rachee Singh, Sharad Agarwal, Matt Calder, Paramvir Bahl | 12m |
Paper proposes Cascara, an online cloud-edge traffic-engineering controller that uses IPFIX flow records, BMP route data, and peering rates to solve a… | ||
| A Social Network Under Social Distancing: Risk-Driven Backbone Management During COVID-19 | Yiting Xia, Ying Zhang, Zhizhen Zhong, Guanqing Yan, Chiun Lin Lim, Satyajeet Singh Ahuja, Soshant Bali, Alexander Nikolaidis, Kimia Ghobadi, Manya Ghobadi | 14m |
Facebook describes a production Risk Simulation System (RSS) that runs year-long failure simulations to quantify backbone risk, modeled with non‑exponential… | ||
| Staying Alive: Connection Path Reselection at the Edge | Raul Landa, Lorenzo Saino, Lennert Buytenhek, Joao Taveira Araujo | 12m |
Connection Path Reselection (CPR) augments TCP at Fastly edge POPs to detect stalled connections by monitoring transport-layer forward progress and to reroute… | ||
| Debugging Transient Faults in Data Centers using Synchronized Network-wide Packet | Pravein Govindan Kannan, Nishant Budhdev, Raj Joshi, Mun Choon Chan | 12m |
Implements SyNDB: programmable-switch packet recording with nanosecond-synced data-plane clocks, ring-buffered compressed “p-records,” and control-plane… | ||
| Alohamora: Reviving HTTP/2 Push and Preload by Adapting Policies On the Fly | Nikhil Kansal, Murali Ramanujam, Ravi Netravali | 12m |
Alohomora uses reinforcement learning to pick per-load HTTP/2 push and preload actions, trains one model per cluster of structurally similar pages, and… | ||
| SENSEI: Aligning Video Streaming Quality with Dynamic User Sensitivity | Xu Zhang, Yiyang Ou, Siddhartha Sen, Junchen Jiang | 13m |
Shows that per-video crowdsourced sensitivity profiles let adaptation trade short-term rebuffering for higher quality at important moments, implemented in a… | ||
| GAIA: A System for Interactive Analysis on Distributed Graphs Using a High-Level Language | Zhengping Qian, Chenqiang Min, Longbin Lai, Yong Fang, Gaofeng Li, Youyang Yao, Bingqing Lyu, Xiaoli Zhou, Zhimin Chen, Jingren Zhou | 14m |
GAIA compiles Gremlin into a distributed dataflow and ships a runtime that automatically parallelizes nested, dynamic traversals. It introduces the… | ||
| TEGRA: Efficient Ad-Hoc Analytics on Evolving Graphs | Anand Padmanabha Iyer, Qifan Pu, Kishan Patel, Joseph E. Gonzalez, Ion Stoica | 12m |
TEGRA exposes evolving graphs via a Timelapse API and stores versions with a persistent adaptive radix tree, enabling snapshot reads without replaying deltas… | ||
| Unifying Timestamp with Transaction Ordering for MVCC with Decentralized Scalar Timestamp | Xingda Wei, Rong Chen, Haibo Chen, Zhaoguo Wang, Zhenhan Gong, Binyu Zang | 11m |
Proposes decentralized scalar timestamp DST that removes a centralized timestamp service while remaining a single-word scalar for storage and network… | ||
| When to Hedge in Interactive Services | Mia Primorac, Katerina Argyraki, Edouard Bugnion | 13m |
Compares hedging vs optimal load balancing, shows hedging often gives no benefit above ~60% utilization and can cause congestion, defines an idealized hedge… | ||
| Move Fast and Meet Deadlines: Fine-grained Real-time Stream Processing with Cameo | Le Xu, Shivaram Venkataraman, Indranil Gupta, Luo Mai, Rahul Potharaju | 12m |
Cameo treats operator invocations as serverless functions and attaches priority context metadata to events so schedulers stay stateless while propagating… | ||
| Whiz: Data-Driven Analytics Execution | Robert Grandl, Arjun Singhvi, Raajay Viswanathan, Aditya Akella | 11m |
A prototype called Whiz cleanly separates computation from intermediate data and adds a data service that collects runtime stats, applies execution and… | ||
| Pushing the Physical Limits of IoT Devices with Programmable Metasurfaces | Lili Chen, Wenjun Hu, Kyle Jamieson, Xiaojiang Chen, Dingyi Fang, Jeremy Gummeson | 13m |
LLAMA builds a programmable metasurface for the 2.4 GHz ISM band that rotates polarization to fix mismatch between cheap single-antenna IoT devices. A 48×48 cm… | ||
| Bootstrapping Battery-free Wireless Networks: Efficient Neighbor Discovery and | Kai Geissdoerfer, Marco Zimmerling | 10m |
Presents Find, a neighbor-discovery protocol that inserts geometric waiting (randomized wake delays) to break persistent wake-up offsets, and Flync, a… | ||
| AIRCODE: Hidden Screen-Camera Communication on an Invisible and Inaudible Dual Channel | Kun Qian, Yumeng Lu, Zheng Yang, Kai Zhang, Kehong Huang, Xinjun Cai, Chenshu Wu, Yunhao Liu | 11m |
AIRCODE achieves imperceptible screen-camera transmission at about 1 Mbps with reported BER ≈ 5% by splitting payload and metadata: video frames carry… | ||
| Device-Based LTE Latency Reduction at the Application Layer | Zhaowei Tan, Jinghao Zhao, Yuanjie Li, Yifei Xu, Songwu Lu | 12m |
LRP is a user‑space Android solution that cuts LTE uplink latency by up to 7.4x (median 42ms → 5ms) by eliminating two measured bottlenecks: DRX dose latency… | ||
| BMC: Accelerating Memcached using Safe In-kernel Caching and Pre-stack Processing | Yoann Ghigoff, Julien Sopena, Kahina Lazri, Antoine Blin, Gilles Muller | 13m |
BMC implements an in-kernel Memcached cache that processes UDP get requests at NIC interrupt time using eBPF/XDP, splitting logic into seven eBPF programs and… | ||
| Segcache: a memory-efficient and scalable in-memory key-value cache for small objects | Juncheng Yang, Yao Yue, Rashmi Vinayak | 12m |
Segment-structured cache that groups small objects by similar creation/expiration into fixed-size segments, lifts most per-object metadata into shared segment… | ||
| When Cloud Storage Meets RDMA | Yixiao Gao, Qiang Li, Lingbo Tang, Yongqing Xi, Pengcheng Zhang, Wenwen Peng, Bo Li, Yaohui Wu, Shaozong Liu, Lei Yan, Fei Feng, Yan Zhuang, Fan Liu, Pan Liu, Xingkui Liu, Zhongjie Wu, Junping Wu, Zheng Cao, Chen Tian, Jinbo Wu, Jiaji Zhu, Haiyong Wang, Dennis Cai, Jiesheng Wu | 11m |
Pangu integrates RDMA into production cloud storage and reports concrete wins: a user‑space stack (USS OS and USS file system) boosts CPU efficiency by over… | ||
| Prism: Proxies without the Pain | Yutaro Hayakawa, Michio Honda, Douglas Santry, Lars Eggert | 12m |
Prism moves established connections from front-end to backend using TCP handoff and a two-phase handoff that blocks client packets while the switch rules… | ||
| Programming Network Stack for Middleboxes with Rubik | Hao Li, Changhao Wu, Guangda Sun, Peng Zhang, Danfeng Shan, Tian Pan, Chengchen Hu | 12m |
Rubik is a Python-based DSL plus compiler that generates optimized C network stacks for middleboxes, using a domain-specific IR to cluster and eliminate… | ||
| Flightplan: Dataplane Disaggregation and Placement for P4 Programs | Nik Sultana, John Sonchack, Hans Giesen, Isaac Pedisich, Zhaoyang Han, Nishanth Shyamkumar, Shivani Burad, André DeHon, Boon Thau Loo | 12m |
Flightplan automatically splits a monolithic P4 program into cooperating subprograms, generates resource dependency rules from code, and uses a planner-based… | ||
| MilliSort and MilliQuery: Large-Scale Data-Intensive Computing in Milliseconds | Yilong Li, Seo Jin Park, John Ousterhout | 12m |
Benchmarking shows flash-burst analytics are feasible: MilliSort sorts 0.84M records in 1 ms using 120 servers by using recursive hierarchical partitioning to… | ||
| EPaxos Revisited | Sarah Tollman, Seo Jin Park, John Ousterhout | 12m |
Re-evaluation reproduces EPaxos’s low median commit latency but shows 99th-percentile latency often worse than Multi‑Paxos — sometimes more than 4× worse — and… | ||
| Ship Compute or Ship Data? Why Not Both? | Jie You, Jingfeng Wu, Xin Jin, Mosharaf Chowdhury | 13m |
Kayak is a prototype that adaptively splits requests between a KV (ship data) path and an RPC (ship compute) path by tuning rate limit and request fraction… | ||
| Caerus: NIMBLE Task Scheduling for Serverless Analytics | Hong Zhang, Yupeng Tang, Anurag Khandelwal, Jingrong Chen, Ion Stoica | 12m |
Caerus is a task-level scheduler that implements the NIMBLE algorithm to schedule serverless analytics tasks by computing per-task launch times. It models… | ||
| Ownership: A Distributed Futures System for Fine-Grained Tasks | Stephanie Wang, Eric Liang, Edward Oakes, Ben Hindman, Frank Sifei Luan, Audrey Cheng, Ion Stoica | 12m |
Ownership claims a decentralized runtime that shards metadata by object owner, trades global persistence for simple recovery, and achieves 1ms latency per task… | ||
| Fault-Tolerant Replication with Pull-Based Consensus in MongoDB | Siyuan Zhou, Shuai Mu | 12m |
Claims linearizable replication that tolerates any minority failure by adapting Raft into a novel pull-based data synchronization model where any replica can… | ||
| Mistify: Automating DNN Model Porting for On-Device Inference at the Edge | Peizhen Guo, Bo Hu, Wenjun Hu | 12m |
Mistify builds an automated server-side pipeline that converts a cloud model into many edge-ready variants by exposing a single configuration file and running… | ||
| Elastic Resource Sharing for Distributed Deep Learning | Changho Hwang, Taehyun Kim, Sunghyun Kim, Jinwoo Shin, KyoungSoo Park | 12m |
Proposes Apathetic Future Share (AFS), an elastic resource‑sharing algorithm that balances short‑job prioritization and per‑job scaling efficiency, and shows… | ||
| ATP: In-network Aggregation for Multi-tenant Learning | ChonLam Lao, Yanfang Le, Kshiteej Mahajan, Yixi Chen, Wenfei Wu, Aditya Akella, Michael Swift | 12m |
Claims up to 38–66% training throughput improvement for multi-tenant distributed DNN jobs by moving gradient aggregation into programmable switches. Implements… | ||
| On the Use of ML for Blackbox System Performance Prediction | Silvery Fu, Saurabh Gupta, Radhika Mittal, Sylvia Ratnasamy | 12m |
Evaluates six ML models on 13 real applications and finds 12 of 13 show inherent variability that limits prediction accuracy, with the oracle error exceeding… | ||
| Scaling Distributed Machine Learning with In-Network Aggregation | Amedeo Sapio, Marco Canini, Chen-Yu Ho, Jacob Nelson, Panos Kalnis, Changhoon Kim, Arvind Krishnamurthy, Masoud Moshref, Dan R. K. Ports, Peter Richtarik | 11m |
Implements SwitchML: a co-designed system that pushes aggregation into a programmable switch dataplane to reduce model-update volume and offload the all-reduce… | ||
| Efficient Wideband Spectrum Sensing Using MEMS Acoustic Resonators | Junfeng Guan, Jitian Zhang, Ruochen Lu, Hyungjoo Seo, Jin Zhou, Songbin Gong, Haitham Hassanieh | 13m |
System samples along the frequency axis using a MEMS acoustic spike-train RF filter with 19 spikes spaced 22 MHz apart to cover 418 MHz, then recovers band… | ||
| WiForce: Wireless Sensing and Localization of Contact Forces on a Space Continuum | Agrim Gupta, Cédric Girerd, Manideep Dunna, Qiming Zhang, Raghav Subbaraman, Tania Morimoto, Dinesh Bharadia | 12m |
Wireless WiForce converts contact force magnitude and location into phase shifts on a backscatter tag and reads them wirelessly using an OFDM reader and… | ||
| MAVL: Multiresolution Analysis of Voice Localization | Mei Wang, Wei Sun, Lili Qiu | 12m |
MAVL localizes a user with a single smart speaker and an eight-microphone uniform circular array by (1) separating coherent multipath AoAs via multi-resolution… | ||
| From Conception to Retirement: a Lifetime Story of a 3-Year-Old Wireless Beacon System | Yi Ding, Ling Liu, Yang Yu, Yunhuai Liu, Desheng Zhang, Tian He | 12m |
Three-year city-scale deployment of customized BLE beacons (pilot 8 months, operation 28 months) places a device at each merchant to infer arrival/departure… | ||
| EarFisher: Detecting Wireless Eavesdroppers by Stimulating and Sensing Memory EMR | Cheng Shen, Jun Huang | 12m |
System induces bait web-page bursts and uses a BladeRF-based detector to sense memory EMR peaks produced when a receiver writes every packet to RAM, separating… | ||
| Adapting Wireless Mesh Network Configuration from Simulation to Reality via Deep Learning | Junyang Shi, Mo Sha, Xi Peng | 12m |
A system trains a teacher-student neural network using large-scale simulator outputs plus a few field measurements to close the simulation-to-reality gap with… | ||
| Practical Null Steering in Millimeter Wave Networks | Sohrab Madani, Suraj Jog, Jesus O. Lacruz, Joerg Widmer, Haitham Hassanieh | 13m |
Nulli‑Fi creates directional nulls on commercial 60 GHz, 16‑element phased arrays (IEEE 802.11ad) using a provably optimal phase‑pair algorithm plus a discrete… | ||
| SyncScatter: Enabling WiFi like synchronization and range for WiFi backscatter | Manideep Dunna, Miao Meng, Po-Han Wang, Chi Zhang, Patrick Mercier, Dinesh Bharadia | 12m |
SyncScatter demonstrates accurate synchronization to incident WiFi signals and sensitivity pushed to the theoretical backscatter link budget, backed by a… | ||
| Verification and Redesign of OFDM Backscatter | Xin Liu, Zicheng Chi, Wei Wang, Yao Yao, Pei Hao, Ting Zhu | 13m |
Identifies why OFDM backscatter fails: WiFi receiver phase error correction cancels symbol-level tag phase shifts, breaking codeword-translation designs… | ||
| Simplifying Backscatter Deployment: Full-Duplex LoRa Backscatter | Mohamad Katanbaf, Anthony Weinand, Vamsi Talla | 12m |
Claims a working long-range full‑duplex LoRa backscatter reader that achieves 78 dB self‑interference cancellation using a hybrid coupler, a microcontroller… | ||
| One Protocol to Rule Them All: Wireless Network-on-Chip using Deep Reinforcement Learning | Suraj Jog, Zikun Liu, Antonio Franques, Vimuth Fernando, Sergi Abadal, Josep Torrellas, Haitham Hassanieh | 13m |
Neural-MCP builds adaptive medium-access protocols for Wireless Network-on-Chip using deep reinforcement learning to learn traffic structure, correlations, and… | ||
| LightGuardian: A Full-Visibility, Lightweight, In-band Telemetry System Using Sketchlets | Yikai Zhao, Kaicheng Yang, Zirui Liu, Tong Yang, Li Chen, Shiyi Liu, Naiqian Zheng, Ruixin Wang, Hanbo Wu, Yi Wang, Nicholas Zhang | 12m |
NSDI '21 - LightGuardian: A Full-Visibility, Lightweight, In-band Telemetry System Using Sketchlets
Yikai Zhao, Kaicheng Yang, and Zirui Liu, Peking… | ||
| Fast and Light Bandwidth Testing for Internet Users | Xinlei Yang, Xianlong Wang, Zhenhua Li, Yunhao Liu, Feng Qian, Liangyi Gong, Rui Miao, Tianyin Xu | 11m |
FastBTS implements fuzzy rejection sampling with a real-time crucial interval detector and elastic bandwidth probing, plus data-driven server selection and… | ||
| Toward Nearly-Zero-Error Sketching via Compressive Sensing | Qun Huang, Siyuan Sheng, Xiang Chen, Yungang Bao, Rui Zhang, Yanwei Xu, Gong Zhang | 12m |
Claims near-zero per-flow error for almost all flows by combining compressive sensing with sketch structures, producing two algorithms called SiX sketch and… | ||
| nothing matches | ||