FAST '23: 21st USENIX Conference on File and Storage Technologies
2023-02-21..2023-02-23
Santa Clara, United States
as the organisers put it
21st USENIX Conference on File and Storage Technologies (FAST '23), February 21–23, 2023, Hyatt Regency Santa Clara, Santa Clara, CA, USA. FAST brings together storage-system researchers and practitioners to explore new directions in the design, implementation, evaluation, and deployment of storage systems.
who it's for
storage engineers, distributed systems engineers, file system developers, cloud infrastructure engineers, site reliability engineers, machine learning systems engineers
topics
erasure coding with locally recoverable codes; MSR code sub-packetization repair; cloud deduplication and tiering; fail-slow drive detection; log-structured key-value stores; memory-disaggregated RDMA key-value stores; learned indexes for RDMA systems; learned caching for deep learning; concurrent journaling for ext4; computational storage unified IO stack
| title | author | length |
|---|---|---|
| Building and Operating a Pretty Big Storage System (My Adventures in Amazon S3) | Andy Warfield | 53m |
Describes building S3 at scale with concrete numbers and engineering choices: 31 regions and 99 AZs, ~280 trillion objects, 100M requests/sec and 400 Tb/s… | ||
| Practical Design Considerations for Wide Locally Recoverable Codes (LRCs) | Saurabh Kadekodi, Shashwat Silas, David Clausen, Arif Merchant | 17m |
Designs produce Uniform Cauchy LRCs built by expanding Azure LRC local groups to cover parity chunks using Cauchy coefficients, constructible from existing… | ||
| ParaRC: Embracing Sub-Packetization for Repair Parallelization in MSR-Coded Storage | Xiaolu Li, Keyun Cheng, Kaicheng Tang, Patrick P. C. Lee, Yuchong Hu, Dan Feng, Jie Li, Ting-Yi Wu | 17m |
ParaRC leverages sub-packetization of MSR codes to parallelize sub-block repairs, models the problem as a colored graph called a PEC DAG to expose the… | ||
| InftyDedup: Scalable and Cost-Effective Cloud Tiering with Deduplication | Iwona Kotlarska, Andrzej Jackowski, Krzysztof Lichota, Michal Welnicki, Cezary Dubnicki, Konrad Iwanicki | 14m |
A cloud-native system implements deduplication entirely in the cloud, using distributed batch jobs and spot instances to process backup metadata (file recipes… | ||
| Perseus: A Fail-Slow Detection Framework for Cloud Storage Systems | Ruiming Lu, Erci Xu, Yiming Zhang, Fengyi Zhu, Zhaosheng Zhu, Mengtian Wang, Zongpeng Zhu, Guangtao Xue, Jiwu Shu, Minglu Li, Jiesheng Wu | 16m |
Perseus detects fail-slow drives non‑intrusively by modeling latency versus throughput with a light regression model, using outlier removal, an adaptive upper… | ||
| ADOC: Automatically Harmonizing Dataflow Between Components in Log-Structured Key-Value | Jinghuan Yu, Sam H. Noh, Young-ri Choi, Chun Jason Xue | 15m |
Claims that LSM write stalls come from a universal data overflow — imbalanced input vs output across memtable, L0 and deeper compactions — and not any single… | ||
| FUSEE: A Fully Memory-Disaggregated Key-Value Store | Jiacheng Shen, Pengfei Zuo, Xuchuan Luo, Tianyi Yang, Yuxin Su, Yangfan Zhou, Michael R. Lyu | 16m |
FUSEE builds a fully memory-disaggregated key-value store that moves index and memory-management metadata onto memory nodes and lets clients manage them… | ||
| ROLEX: A Scalable RDMA-oriented Learned Key-Value Store for Disaggregated Memory Systems | Pengfei Li, Yu Hua, Pengfei Zuo, Zhangyu Chen, Jiajie Sheng | 15m |
ROLEX is a key-value store that pairs retraining-decoupled learned index models with RDMA so compute nodes perform updates via one-sided RDMA verbs while… | ||
| GL-Cache: Group-level learning for efficient and high-performance caching | Juncheng Yang, Ziming Mao, Yao Yue, K.V. Rashmi | 16m |
Groups objects by insertion-time grouping and trains a model on group snapshots (seven features, using gradient-boosted trees) so inference and state are… | ||
| SHADE: Enable Fundamental Cacheability for Distributed Deep Learning Training | Redwan Ibne Seraj Khan, Ahmad Hossein Yazdani, Yuqi Fu, Arnab K. Paul, Bo Ji, Xun Jian, Yue Cheng, Ali R. Butt | 18m |
SHADE is a DLT-aware cache that computes per-sample importance via loss decomposition and rank-based importance scores, applies a patch sampling policy, and… | ||
| Intelligent Resource Scheduling for Co-located Latency-critical Services: A Multi-Model | Lei Liu, Xinglei Dou, Yuetao Chen | 16m |
Claims an ML scheduler, OSML, that detects resource cliffs and finds the optimal allocation area using three collaborative models (Model A predicts cliffs/OA… | ||
| CJFS: Concurrent Journaling for Better Scalability | Joontaek Oh, Seung Won Yoo, Hojin Nam, Changwoo Min, Youjip Won | 15m |
Extends EXT4 to allow concurrent journal commits by building a patched filesystem (CJFS) that implements Dual Thread Journaling, Multi-version Shadow Paging… | ||
| Unsafe at Any Copy: Name Collisions from Mixing Case Sensitivities | Aditya Basu, John Sampson, Zhiyun Qian, Trent Jaeger | 15m |
Work builds an auditd-based detector and a test suite that drives copy utilities (cp, rsync, tar) and reproduces the Git CVE; experiments on ext4 with… | ||
| ConfD: Analyzing Configuration Dependencies of File Systems for Fun and Profit | Tabassum Mahmud, Om Rameshwar Gatla, Duo Zhang, Carson Love, Ryan Bumann, Mai Zheng | 19m |
Claims that an extensible tool called ConfD automatically extracts multi-level configuration dependencies and finds real issues: ConfD extracts 154… | ||
| HadaFS: A File System Bridging the Local and Shared Burst Buffer for Exascale | Xiaobin He, Bin Yang, Jie Gao, Wei Xiao, Qi Chen, Shupeng Shi, Dexun Chen, Weiguo Liu, Wei Xue, Zuo-ning Chen | 16m |
HadaFS is a deployed burst-buffer file system that bridges local and shared BBs by implementing a Localized Triage Architecture to bind clients to selected… | ||
| Fisc: A Large-scale Cloud-native-oriented File System | Qiang Li, Lulu Chen, Xiaoliang Wang, Shuo Huang, Qiao Xiang, Yuanyuan Dong, Wenhui Yao, Minfei Huang, Puyuan Yang, Shanyang Liu, Zhaosheng Zhu, Huayong Wang, Haonan Qiu, Derui Liu, Shaozong Liu, Yujie Zhou, Yaohui Wu, Zhiwu Wu, Shang Gao, Chao Han, Zicheng Luo, Yuchao Shao, Gexiao Tian, Zhongjie Wu, Zheng Cao, Jinbo Wu, Jiwu Shu, Jie Wu, Jiesheng Wu, Junping Wu, Jiaji Zhu | 17m |
A cloud-native file system that offloads heavy client functions into local agents and DPU-based virtio devices and introduces vrpc and a file-based range table… | ||
| TENET: Memory Safe and Fault Tolerant Persistent Transactional Memory | R. Madhava Krishnan, Diyu Zhou, Wook-Hee Kim, Sudarsun Kannan, Sanidhya Kashyap, Changwoo Min | 21m |
TENET is a PTM framework built on TimeStone that prevents software and media corruption by making NVM read‑only to applications via Memory Protection Keys… | ||
| MadFS: Per-File Virtualization for Userspace Persistent Memory Filesystems | Shawn (Wanxiang) Zhong, Chenhao Ye, Guanzhou Hu, Suyan Qu, Andrea C. Arpaci-Dusseau, Remzi H. Arpaci-Dusseau, Michael Swift | 16m |
MadFS implements per-file virtualization and metadata embedding to move block maps, virtual size and a compact 8‑byte metadata log into file data so userspace… | ||
| On Stacking a Persistent Memory File System on Legacy File Systems | Hobin Woo, Daegyu Han, Seungjoon Ha, Sam H. Noh, Beomseok Nam | 18m |
SPFS stacks a persistent memory layer on top of legacy file systems (ext4, f2fs, xfs) and exposes NVMM (Intel DCPMM, MBDM) as a persistent writeback cache that… | ||
| Citron: Distributed Range Lock Management with One-sided RDMA | Jian Gao, Youyou Lu, Minhui Xie, Qing Wang, Jiwu Shu | 15m |
Citron implements a distributed range lock manager that uses one-sided RDMA and a static segment tree layout so clients bypass server-side CPUs. It maps any… | ||
| Patronus: High-Performance and Protective Remote Memory | Bin Yan, Youyou Lu, Qing Wang, Minhui Xie, Jiwu Shu | 16m |
PATRONUS implements a protective RDMA remote-memory system that preserves one-sided access while using memory windows and tokenized permissions to enforce… | ||
| More Than Capacity: Performance-oriented Evolution of Pangu in Alibaba | Qiang Li, Qiao Xiang, Yuxin Wang, Haohao Song, Ridi Wen, Wenhui Yao, Yuanyuan Dong, Shuqi Zhao, Shuo Huang, Zhaosheng Zhu, Huayong Wang, Shanyang Liu, Lulu Chen, Zhiwu Wu, Haonan Qiu, Derui Liu, Gexiao Tian, Chao Han, Shaozong Liu, Yaohui Wu, Zicheng Luo, Yuchao Shao, Junping Wu, Zheng Cao, Zhongjie Wu, Jiaji Zhu, Jinbo Wu, Jiwu Shu, Jiesheng Wu | 15m |
A global unified storage system evolves to deliver 100‑microsecond average latency by deploying a user‑space storage OS (USS OS), RDMA from 25Gbps to 100Gbps… | ||
| λ-IO: A Unified IO Stack for Computational Storage | Zhe Yang, Youyou Lu, Xiaojian Liao, Youmin Chen, Junru Li, Siyu He, Jiwu Shu | 17m |
λ-IO presents a full-stack software and hardware prototype (OpenSSD acting as NVMe) that routes per-request computation to whichever side is predicted faster… | ||
| Revitalizing the Forgotten On-Chip DMA to Expedite Data Movement in NVM-based Storage | Jingbo Su, Jiahao Li, Luofan Chen, Cheng Li, Kai Zhang, Liang Yang, Sam H. Noh, Yinlong Xu | 18m |
Claims that reviving the on-chip DMA plus CPU cooperation cuts DRAM↔NVM data-movement cost and lowers the I/O size threshold (reads benefit above ~32 KB… | ||
| NVMeVirt: A Versatile Software-defined Virtual NVMe Device | Sang-Hoon Kim, Jaehoon Shim, Euidong Lee, Seongyeop Jeong, Ilkueon Kang, Jin-Soo Kim | 18m |
Kernel module creates a real-looking virtual PCI device via a virtual PCI bus, runs a dispatcher thread that polls doorbells and splits NVMe commands into… | ||
| SMRSTORE: A Storage Engine for Cloud Object Storage on HM-SMR Drives | Su Zhou, Erci Xu, Hao Wu, Yu Du, Jiacheng Cui, Wanyu Fu, Chang Liu, Yingni Wang, Wenbo Wang, Shouqu Sun, Xianfei Wang, Bo Feng, Biyun Zhu, Xin Tong, Weikang Kong, Linyan Liu, Zhongjie Wu, Jinbo Wu, Qingchao Luo, Jiesheng Wu | 15m |
SMRSTORE is a log-structured user-space engine that writes chunk records directly to zoned HM-SMR drives via zoned block I/O, implements per-zone in-memory… | ||
| Multi-view Feature-based SSD Failure Prediction: What, When, and Why | Yuqi Zhang, Wenwen Hao, Ben Niu, Kangkang Liu, Shuyang Wang, Na Liu, Xing He, Yongwong Gwon, Chankyu Koh | 17m |
Proposes MVTRF multi-view multi-task random forest that concatenates per-log attributes with histogram and sequence features extracted from long-term and… | ||
| Fast Application Launch on Personal Computing/Communication Devices | Junhee Ryu, Dongeun Lee, Kang G. Shin, Kyungtae Kang | 18m |
Paralfetch is a Linux kernel prefetcher implemented on desktop/laptop, Raspberry Pi 3 and an Android smartphone that collects accurate tracing of… | ||
| Integrated Host-SSD Mapping Table Management for Improving User Experience of Smartphones | Yoona Kim, Inhyuk Choi, Juhyung Park, Jaeheon Lee, Sungjin Lee, Jihong Kim | 20m |
Prototype HPBvalve integrates foreground app-centric caching with dynamic HPB resizing to reduce L2P miss penalties on a Snapdragon 888 dev board (12 GB DRAM… | ||
| nothing matches | ||