USENIX ATC '19
2019-07-10..2019-07-12
Renton, United States
as the organisers put it
2019 USENIX Annual Technical Conference will take place July 10–12, 2019, at the Hyatt Regency Lake Washington in Renton, WA, USA. USENIX ATC '19 will bring together leading systems researchers for cutting-edge systems research and the opportunity to gain insight into a wealth of must-know topics.
| title | author | length |
|---|---|---|
| Measure, Then Build | Remzi H. Arpaci-Dusseau | 01h 00m |
Advocates an iterative measure, then build workflow: profile real systems to find surprising bottlenecks, then redesign. Examples include measuring existing… | ||
| The Design and Operation of CloudLab | Dmitry Duplyakin, Robert Ricci, Aleksander Maricq, Gary Wong, Jonathon Duerig, Eric Eide, Leigh Stoller, Mike Hibler, David Johnson, Kirk Webb, Aditya Akella, Kuangching Wang, Glenn Ricart, Larry Landweber, Chip Elliott, Michael Zink, Emmanuel Cecchet, Snigdhaswin Kar, Prabodh Mishra | 19m |
CloudLab is a distributed hardware testbed spanning Utah Wisconsin and Clemson that serves 4,000 users who run over 79,000 experiments on 2,250 servers and… | ||
| Everyone Loves File: File Storage Service (FSS) in Oracle Cloud Infrastructure | Bradley C. Kuszmaul, Matteo Frigo, Justin Mazzola Paluska | 22m |
Oracle FSS implements a managed NFS by building a pipelined Paxos replicated backend: a scalable block store offering linearizable multipage limited-size… | ||
| Zanzibar: Google’s Consistent, Global Authorization System | Ruoming Pang, Ramon Caceres, Mike Burrows, Zhifeng Chen, Pratik Dave, Nathan Germer, Alexander Golynski, Kevin Graney, Nina Kang, Lea Kissner, Jeffrey L. Korn, Abhishek Parmar, Christina D. Richards, Mengzhi Wang | 21m |
Global authorization system that stores and evaluates access control lists for hundreds of services, implemented on Spanner and a custom in-memory index, and… | ||
| IASO: A Fail-Slow Detection and Mitigation Framework for Distributed Storage Services | Biswaranjan Panda, Deepthi Srinivasan, Huan Ke, Karan Gupta, Vinayak Khot, Haryadi S. Gunawi | 24m |
System detects and isolates “fail‑slow” nodes using AIMD-inspired scoring on per‑peer timeout ratios, publishing per‑minute scores to a score DB and taking the… | ||
| PARTISAN: Scaling the Distributed Actor Runtime | Christopher S. Meiklejohn, Heather Miller, Peter Alvaro | 22m |
PARTISAN is an Erlang library that replaces Distributed Erlang’s single full‑mesh, single‑TCP assumptions with runtime‑selectable overlays (full‑mesh… | ||
| Unleashing the Power of Learning: An Enhanced Learning-Based Approach for Dynamic Binary Translation | Changheng Song, Wenwen Wang, Pen-Chung Yew, Antonia Zhai, Weihua Zhang | 20m |
Enhanced learning extracts translation rules from paired guest/host binaries, uses symbolic execution + SMT-based equivalence checking, and integrates learned… | ||
| Transactuations: Where Transactions Meet the Physical World | Aritra Sengupta, Tanakorn Leesatapornwongsa, Masoud Saeida Ardekani, Cesar A. Stuardo | 21m |
Identifies that smart‑home apps suffer irreversible mismatches between device and software state and proposes transactuation, an abstraction that enforces a… | ||
| Not So Fast: Analyzing the Performance of WebAssembly vs. Native Code | Abhinav Jangda, Bobby Powers, Emery D. Berger, Arjun Guha | 20m |
Built Browsix-Wasm to run unmodified Unix programs in the browser and benchmarked them using SPEC CPU; WebAssembly binaries run about 45% slower on Firefox and… | ||
| Extension Framework for File Systems in User space | Ashish Bijlani, Umakishore Ramachandran | 19m |
ExtFUSE implements eBPF kernel extensions that let FUSE file systems register thin in-kernel handlers for metadata caching, permission checks and direct I/O… | ||
| FlexGroup Volumes: A Distributed WAFL File System | Ram Kesavan, Jason Hennessey, Richard Jernigan, Peter Macko, Keith A. Smith, Daniel Tennant, Bharadwaj V. R. | 19m |
FlexGroup turns many local WAFL volumes into a single distributed namespace by adding a remote links layer, a route cache with delegations for consistency, and… | ||
| EROFS: A Compression-friendly Readonly File System for Resource-scarce Devices | Xiang Gao, Mingkai Dong, Xie Miao, Wei Du, Chao Yu, Haibo Chen | 17m |
EROFS implements fixed-sized output compression and multiple decompression modes including in-place decompression and small per-CPU buffer decompression; it… | ||
| QZFS: QAT Accelerated Compression in File System for Application Agnostic and Cost Efficient Data Storage | Xiaokang Hu, Fuzong Wang, Weigang Li, Jian Li, Haibing Guan | 22m |
Integration of Intel® QAT ASIC into ZFS accelerates gzip (de)compression at the filesystem layer, adds a compression service engine that selects hardware or… | ||
| Apache Nemo: A Framework for Building Distributed Dataflow Optimization Policies | Youngseok Yang, Jeongyoon Eo, Geon-Woo Kim, Joo Yeon Kim, Sanha Lee, Jangho Seo, Won Wook Song, Byung-Gon Chun | 17m |
Apache Nemo is an open-source Java framework (≈32K LOC) that converts jobs from Spark or Beam into an IR DAG and applies compile-time and runtime optimization… | ||
| Tangram: Bridging Immutable and Mutable Abstractions for Distributed Data Analytics | Yuzhen Huang, Xiao Yan, Guanxian Jiang, Tatiana Jin, James Cheng, An Xu, Zhanhao Liu, Shuo Tu | 21m |
Tangram implements MapUpdate to infer whether collections are mutable or immutable and lets update functions perform in-place, key-value updates while map… | ||
| STRADS-AP: Simplifying Distributed Machine Learning Programming without Introducing a New Programming Model | Jin Kyu Kim, Abutalib Aghayev, Garth A. Gibson, Eric P. Xing | 19m |
STRADS-AP automatically parallelizes ordinary sequential ML code so that converting published algorithms to a cluster requires few mechanical changes rather… | ||
| SOPHIA: Online Reconfiguration of Clustered NoSQL Databases for Time-Varying Workloads | Ashraf Mahgoub, Paul Wood, Alexander Michaelson Medoff, Subrata Mitra, Folker Meyer, Somali Chaterji, Saurabh Bagchi | 18m |
SOPHIA automatically plans online NoSQL reconfigurations by combining a Rafiki static tuner to pick impactful parameters, a first-order Markov predictor for… | ||
| DistCache: Provable Load Balancing for Large-Scale Storage Systems with Distributed Caching | Zaoxing Liu, Zhihao Bai, Zhenming Liu, Xiaozhou Li, Changhoon Kim, Vladimir Braverman, Xin Jin, Ion Stoica | 21m |
DistCache splits a single hot-cache into many small nodes, allocating hot items with two independent hash functions and routing queries with the… | ||
| Protocol-Aware Recovery for Consensus-Based Storage | Ramnatthan Alagappan, Aishwarya Ganesan, Eric Lee, Aws Albarghouthi, Vijay Chidambaram, Andrea C. Arpaci-Dusseau, Remzi H. Arpaci-Dusseau | 19m |
Paper introduces protocol-aware recovery and implements CTRL to exploit RSM protocol rules to recover from disk corruption. Implementation in LogCabin and… | ||
| Orca: Differential Bug Localization in Large-Scale Services | Ranjita Bhagwan, Rahul Kumar, Chandra Sekhar Maddila, Adithya Abraham Philip | 18m |
Orca takes symptom text and build number, opens the build graph to collect sibling/parent builds, and converts commits to AST diffs and touched paths. It… | ||
| libmpk: Software Abstraction for Intel Memory Protection Keys (Intel MPK) | Soyeon Park, Sangho Lee, Wen Xu, Hyungon Moon, Taesoo Kim | 20m |
libmpk virtualizes Intel MPK so 16 hardware keys become an unlimited virtual key space, prevents protection-key-use-after-free by managing and protecting… | ||
| Effective Static Analysis of Concurrency Use-After-Free Bugs in Linux Device Drivers | Jia-Ju Bai, Julia Lawall, Qiu-Liang Chen, Shi-Min Hu | 15m |
Paper proposes DCUAF, a static analysis that combines local-global analysis to identify driver interface pairs that can run concurrently and then performs a… | ||
| LXDs: Towards Isolation of Kernel Subsystems | Vikram Narayanan, Abhiram Balasubramanian, Charlie Jacobsen, Sarah Spall, Scott Bauer, Michael Quigley, Aftab Hussain, Abdullah Younis, Junjie Shen, Moinak Bhattacharyya, Anton Burtsev | 22m |
A framework called lightweight execution domains isolates legacy kernel subsystems by auto-generating glue from a new IDL and projections, running isolated… | ||
| LegoOS: A Disseminated, Distributed OS for Hardware Resource Disaggregation | Yizhou Shan, Yutong Huang, Yilun Chen, Yiying Zhang | 26m |
LegoOS is a from‑scratch OS (≈200K LOC) that implements a splitkernel of loosely coupled monitors, separates processor (P) and memory (M) components across the… | ||
| JumpSwitches: Restoring the Performance of Indirect Branches In the Era of Spectre | Nadav Amit, Fred Jacobs, Michael Wei | 20m |
JumpSwitches implement just-in-time promotion of indirect calls by runtime learning, an efficient patching worker thread, and on-the-fly binary rewriting… | ||
| Darwin: A Genomics Co-processor Provides up to 15,000X Acceleration on Long Read Assembly | Yatish Turakhia, Gill Bejerano, William J. Dally | 19m |
Darwin: A Genomics Co-processor Provides up to 15,000X Acceleration on Long Read Assembly
Yatish Turakhia and Gill Bejerano, Stanford University; William J… | ||
| The Semantics of Transactions and Weak Memory in x86, Power, ARM, and C++ | Nathan Chong, Tyler Sorensen, John Wickerson | 18m |
Extends existing axiomatic models for x86, Power, ARMv8 and C++ with concrete TM rules and proves sanity properties (transactions never enable previously… | ||
| Who Left Open the Cookie Jar? A Comprehensive Evaluation of Third-Party Cookie Policies | Gertjan Franken, Tom Van Goethem, Wouter Joosen | 27m |
An automated test framework generates hundreds of web scenarios and applies them to 7 browsers and 46 extensions, producing a register of requests via a proxy… | ||
| Multi-Queue Fair Queuing | Mohammad Hedayati, Kai Shen, Michael L. Scott, Mike Marty | 20m |
A scalable fair I/O scheduler called MQFQ is implemented in Linux 4.15 and adapts Start-Time Fair Queueing to multi-queue devices by allowing bounded… | ||
| BRAVO—Biased Locking for Reader-Writer Locks | Dave Dice, Alex Kogan | 21m |
BRAVO augments any reader‑writer lock by adding two integer fields to each lock and using a shared 4,000-entry table where readers hash thread id and lock… | ||
| Mitigating Asymmetric Read and Write Costs in Cuckoo Hashing for Storage Systems | Zhangyu Chen, Yuanyuan Sun, Yu Hua, Yuncheng Guo | 20m |
Zhangyu Chen, Yuanyuan Sun, Yu Hua, and Yuncheng Guo, Huazhong University of Science and Technology
In storage systems, cuckoo hash tables have been widely… | ||
| NICA: An Infrastructure for Inline Acceleration of Network Applications | Haggai Eran, Lior Zeno, Maroun Tork, Gabi Malka, Mark Silberstein | 18m |
NICA builds an ikernel abstraction and runtime that runs AFUs on a Mellanox FPGA+ASIC board, prototypes on Mellanox Anova Flex, and integrates with VMA and… | ||
| E3: Energy-Efficient Microservices on SmartNIC-Accelerated Servers | Ming Liu, Simon Peter, Arvind Krishnamurthy, Phitchaya Mangpo Phothilimthana | 19m |
E3 is a microservice execution platform that offloads small, communication‑intensive services onto Cavium LiquidIO SmartNICs and extends Azure Service Fabric… | ||
| INSIDER: Designing In-Storage Computing System for Emerging High-Performance Drive | Zhenyuan Ruan, Tong He, Jason Cong | 17m |
INSIDER places an FPGA-based reconfigurable drive controller inside the drive, implements a separate control plane (permission checks, per-drive scheduling… | ||
| Cognitive SSD: A Deep Learning Engine for In-Storage Data Retrieval | Shengwen Liang, Ying Wang, Youyou Lu, Zhe Yang, Huawei Li, Xiaowei Li | 16m |
Cognitive SSD places a flash‑side deep learning and graph‑search accelerator called DLG-x inside the SSD, exposes functions via NVMe command extension, and… | ||
| SIMD-X: Programming and Processing of Graph Algorithms on GPUs | Hang Liu, H. Howie Huang | 22m |
Presents a GPU graph framework that exposes an Active-Compute-Combine (ACC) API so algorithms are written in tens of lines, uses just-in-time task management… | ||
| LUMOS: Dependency-Driven Disk-based Graph Processing | Keval Vora | 22m |
Keval Vora, Simon Fraser University
Out-of-core graph processing systems are well-optimized to maintain sequential locality on disk and minimize the amount of… | ||
| NeuGraph: Parallel Deep Neural Network Computation on Large Graphs | Lingxiao Ma, Zhi Yang, Youshan Miao, Jilong Xue, Ming Wu, Lidong Zhou, Yafei Dai | 19m |
NeuGraph implements a TensorFlow-layer that fuses graph and dataflow models, adding trunk-based partitioning and a column-by-column scheduler, custom GPU… | ||
| Pre-Select Static Caching and Neighborhood Ordering for BFS-like Algorithms on Disk-based Graph Engines | Eunjae Lee, Junghyun Kim, Keunhak Lim, Sam H. Noh, Jiwon Seo | 20m |
Proposes a pre-selected in-memory static cache called Basc that stores chosen vertices' edge lists throughout traversal and an I/O-cost-driven graph reorder… | ||
| From Laptop to Lambda: Outsourcing Everyday Jobs to Thousands of Transient Functional Containers | Sadjad Fouladi, Francisco Romero, Dan Iter, Qian Li, Shuvo Chatterjee, Christos Kozyrakis, Matei Zaharia, Keith Winstein | 21m |
A toolkit called gg packages programs as content‑hashed lightweight containers called gg thunks and composes them into a linked GG IR to offload entire jobs to… | ||
| Hodor: Intra-Process Isolation for High-Throughput Data Plane Libraries | Mohammad Hedayati, Spyridoula Gravani, Ethan Johnson, John Criswell, Michael L. Scott, Kai Shen, Mike Marty | 22m |
Introduces protected libraries as an OS abstraction that isolates services inside one address space, implemented with a loader, trampolines, and a kernel… | ||
| A Retargetable System-Level DBT Hypervisor | Tom Spink, Harry Wagstaff, Björn Franke | 17m |
Retargetable DBT hypervisor generates guest modules from an architecture description and combines an offline generator + JIT in a bare-metal VM to translate… | ||
| MTS: Bringing Multi-Tenancy to Virtual Networking | Kashyap Thimmaraju, Saad Hermak, Gabor Retvari, Stefan Schmid | 17m |
MTS isolates tenant networks by moving virtual switching into dedicated VMs and connecting tenant VMs via SR‑IOV virtual functions on a Mellanox ConnectX‑4… | ||
| StreamBox-TZ: Secure Stream Analytics at the Edge with TrustZone | Heejin Park, Shuang Zhai, Long Lu, Felix Xiaozhu Lin | 19m |
StreamBox-TZ implements a secure data plane inside ARM TrustZone that encloses analytics state and computation while leaving scheduling in an untrusted control… | ||
| CoSMIX: A Compiler-based System for Secure Memory Instrumentation and Execution in Enclaves | Meni Orenbach, Yan Michalevsky, Christof Fetzer, Mark Silberstein | 19m |
A compiler instruments memory accesses to route regions to pluggable memory stores and a lightweight enclave runtime handles faults without costly enclave… | ||
| Secured Routines: Language-based Construction of Trusted Execution Environments | Adrien Ghosn, James R. Larus, Edouard Bugnion | 15m |
Language extension for Go called GOTEE adds a go secure keyword that compiles selected goroutines and their data into an SGX enclave, using typed deep-copy… | ||
| Supporting Security Sensitive Tenants in a Bare-Metal Cloud | Amin Mosayyebzadeh, Apoorve Mohan, Sahil Tikale, Mania Abdi, Nabil Schear, Trammell Hudson, Charles Munson, Larry Rudolph, Gene Cooperman, Peter Desnoyers, Orran Krieger | 27m |
Claims a bare-metal cloud architecture called SecCloud that lets tenants pick security vs cost vs performance by running isolation, provisioning and… | ||
| Asynchronous I/O Stack: A Low-latency Kernel I/O Stack for Ultra-Low Latency SSDs | Gyusun Lee, Seokha Shin, Wonsuk Song, Tae Jun Ham, Jae W. Lee, Jinkyu Jeong | 19m |
An asynchronous I/O stack replaces synchronous kernel I/O steps with overlapped operations and a small set of engineering changes: per-core DMA-premapped page… | ||
| M³x: Autonomous Accelerators via Context-Enabled Fast-Path Communication | Nils Asmussen, Michael Roitzsch, Hermann Härtig | 20m |
M³x gives fixed-function accelerators autonomous access to OS services by adding a per-unit DTU, an Accelerator Support Module that implements a UNIX-like 5… | ||
| SmartDedup: Optimizing Deduplication for Resource-constrained Devices | Qirui Yang, Runyu Jin, Ming Zhao | 20m |
SmartDedup implements a hybrid deduplication system on EXT4 and F2FS, prototyped on Nexus 5X and Raspberry Pi 3, that combines in-memory + on-disk fingerprint… | ||
| Data Domain Cloud Tier: Backup here, backup there, deduplicated everywhere! | Abhinav Duggal, Fani Jenkins, Philip Shilane, Ramprasad Chinthekindi, Ritesh Shah, Mahesh Kamat | 21m |
Abhinav Duggal, Fani Jenkins, Philip Shilane, Ramprasad Chinthekindi, Ritesh Shah, and Mahesh Kamat, Dell EMC
Data Domain has added a cloud tier capability to… | ||
| GAIA: An OS Page Cache for Heterogeneous Systems | Tanya Brokhman, Pavel Lifshits, Mark Silberstein | 17m |
GAIA integrates GPU memory into the OS page cache by implementing a weakly-consistent page cache and a lazy release consistency protocol inside Linux… | ||
| Transkernel: Bridging Monolithic Kernels to Peripheral Cores | Liwei Guo, Shuang Zhai, Yi Qiao, Felix Xiaozhu Lin | 17m |
Transkernel runs unmodified Linux suspend/resume on a microcontroller-like peripheral core by using cross-ISA dynamic binary translation plus a tiny set of… | ||
| Detecting Asymmetric Application-layer Denial-of-Service Attacks In-Flight with Finelame | Henri Maxime Demoulin, Isaac Pedisich, Nikos Vasilakis, Vincent Liu, Boon Thau Loo, Linh Thi Xuan Phan | 22m |
Finelame instruments the OS with eBPF probes to map request IDs to processing functions and to account malloc and other resource allocations per request… | ||
| SemperOS: A Distributed Capability System | Matthias Hille, Nils Asmussen, Pramod Bhatotia, Hermann Härtig | 20m |
SemperOS implements a multi‑kernel, HW/SW co‑designed distributed capability system that handles cross‑partition delegation and revoke races by adding a… | ||
| Pragh: Locality-preserving Graph Traversal with Split Live Migration | Xiating Xie, Xingda Wei, Rong Chen, Haibo Chen | 17m |
Pragh implements split live migration that moves values only while leaving keys in place, enabling fine-grained, zero-metadata reconfiguration over RDMA. It… | ||
| ElasticBF: Elastic Bloom Filter with Hotness Awareness for Boosting Read Performance in Large Key-Value Stores | Yongkun Li, Chengjin Tian, Fan Guo, Cheng Li, Yinlong Xu | 14m |
ElasticBF: Elastic Bloom Filter with Hotness Awareness for Boosting Read Performance in Large Key-Value Stores
Yongkun Li, Chengjin Tian, Fan Guo, Cheng Li… | ||
| SILK: Preventing Latency Spikes in Log-Structured Merge Key-Value Stores | Oana Balmau, Florin Dinu, Willy Zwaenepoel, Karan Gupta, Ravishankar Chandhiramoorthi, Diego Didona | 19m |
Implements an I/O scheduler inside RocksDB that allocates bandwidth opportunistically, prioritizes lower-level work and preempts higher-level compactions to… | ||
| Unification of Temporary Storage in the NodeKernel Architecture | Patrick Stuedi, Animesh Trivedi, Jonas Pfefferle, Ana Klimovic, Adrian Schuepbach, Bernard Metzler | 29m |
NodeKernel designs a unified temporary-storage layer and ships it as Crail, a metadata‑driven storage with hierarchical namespaces, storage classes (DRAM and… | ||
| Evaluating File System Reliability on Solid State Drives | Shehbaz Jaffer, Stathis Maneas, Andy Hwang, Bianca Schroeder | 19m |
Built a device-mapper injector that intercepts and corrupts or drops block I/O and runs over a thousand injected fault cases against ext4, Btrfs and F2FS… | ||
| Alleviating Garbage Collection Interference Through Spatial Separation in All Flash Arrays | Jaeho Kim, Kwanghyun Lim, Youngdon Jung, Sungjin Lee, Changwoo Min, Sam H. Noh | 18m |
SWAN partitions an all-flash array into spatial separation front-end SSDs that absorb app writes and back-end SSDs that perform GC, implemented in the host… | ||
| Practical Erase Suspension for Modern Low-latency SSDs | Shine Kim, Jonghyun Bae, Hakbeom Jang, Wenjing Jin, Jeonghun Gong, Seungyeon Lee, Tae Jun Ham, Jae W. Lee | 19m |
Shows three practical erase-suspend methods that pause iterative NAND erase only at safe pulse boundaries to avoid added peripheral cost and cell wear; prior… | ||
| Track-based Translation Layers for Interlaced Magnetic Recording | Mohammad Hossein Hajkazemi, Ajay Narayan Kulkarni, Peter Desnoyers, Timothy R Feldman | 19m |
Evaluation shows read-modify-write incurs up to 2.5× write amplification and high latencies on VM block traces, while three device-side track translation… | ||
| R2P2: Making RPCs first-class datacenter citizens | Marios Kogias, George Prekas, Adrien Ghosn, Jonas Fietz, Edouard Bugnion | 21m |
R2P2 is a UDP-based transport that exposes request–response pairs (identified by src IP/port and client-set request ID) so routers can choose targets and let… | ||
| Your Coflow has Many Flows: Sampling them for Fun and Speed | Akshay Jajoo, Y. Charlie Hu, Xiaojun Lin | 21m |
A new scheduler called Philae samples a few flows per coflow (typically sampling 1% of flows or only for coflows with ≥7 flows), pre-schedules those samples at… | ||
| PostMan: Rapidly Mitigating Bursty Traffic by Offloading Packet Processing | Panpan Jin, Jian Guo, Yikai Xiao, Rong Shi, Yipei Niu, Fangming Liu, Chen Qian, Yang Wang | 20m |
System offloads CPU-heavy per-packet work to on-demand middleboxes called helpers that assemble small packets into large ones, replace TCP/IP headers with a… | ||
| Lancet: A self-correcting Latency Measuring Tool | Marios Kogias, Stephen Mallon, Edouard Bugnion | 22m |
Lancet is a self-correcting latency tool that uses NIC hardware timestamping or a busy-polling latency agent, drives experiments with online statistical tests… | ||
| Pangolin: A Fault-Tolerant Persistent Memory Programming Library | Lu Zhang, Steven Swanson | 19m |
Library provides crash consistency plus media- and software-error protection for NVMM by combining replication of metadata with XOR parity row across 16 GB… | ||
| Pisces: A Scalable and Efficient Persistent Transactional Memory | Jinyu Gu, Qianqian Yu, Xiayang Wang, Zhaoguo Wang, Binyu Zang, Haibing Guan, Haibo Chen | 15m |
Pisces implements a dual-version concurrency control with at most two NVM-resident versions per object and a three-stage commit to separate persistence… | ||
| EdgeWise: A Better Stream Processing Engine for the Edge | Xinwei Fu, Talha Ghaffar, James C. Davis, Dongyoon Lee | 18m |
EdgeWise is an SPE built on Apache Storm and evaluated on Raspberry Pi gateways and an IoT benchmark; it replaces one-worker-per-operator with a fixed-size… | ||
| Analysis of Large-Scale Multi-Tenant GPU Clusters for DNN Training Workloads | Myeongjae Jeon, Shivaram Venkataraman, Amar Phanishayee, Junjie Qian, Wencong Xiao, Fan Yang | 19m |
Analysis uses a 75‑day trace from Microsoft Philly (shared by 14 production groups) combining scheduler logs, hardware counters and AI‑engine logs to profile… | ||
| Lessons and Actions: What We Learned from 10K SSD-Related Storage System Failures | Erci Xu, Mai Zheng, Feng Qin, Yikang Xu, Jiesheng Wu | 21m |
Study analyzes about 10K RASR failures across ~500k SSDs over three years in Alibaba Cloud and quantifies causes: passive heating from mixed rack placement… | ||
| Who's Afraid of Uncorrectable Bit Errors? Online Recovery of Flash Errors with Distributed Redundancy | Amy Tai, Andrew Kryczka, Shobhit O. Kanaujia, Kyle Jamieson, Michael J. Freedman, Asaf Cidon | 18m |
Proposes DIRECT — a set of techniques that let devices expose higher bit error rates by using existing replication and coordination: replicate metadata… | ||
| Dayu: Fast and Low-interference Data Recovery in Very-large Storage Systems | Zhufan Wang, Guangyan Zhang, Yang Wang, Qinglin Yang, Jiaji Zhu | 18m |
Paper builds a centralized timeslot-based scheduler that only schedules tasks likely to finish in the slot, combines a greedy source/destination picker with… | ||
| OPTR: Order-Preserving Translation and Recovery Design for SSDs with a Standard Block Device Interface | Yun-Sheng Chang, Ren-Shuo Liu | 18m |
Proposes OPTR, an FTL+recovery design that keeps the standard block device interface and gives three request‑level crash guarantees: ordered durability… | ||
| Optimizing CNN Model Inference on CPUs | Yizhi Liu, Yao Wang, Ruofei Yu, Mu Li, Vin Sharma, Yida Wang | 19m |
NeoCPU replaces third-party kernels with tunable operator templates (block sizes, register count, loop unrolling) and records real-device timings per… | ||
| Accelerating Rule-matching Systems with Learned Rankers | Zhao Lucis Li, Chieh-Jan Mike Liang, Wei Bai, Qiming Zheng, Yongqiang Xiong, Guangzhong Sun | 15m |
A learned ranker classifies which regex rules to try first, implemented as a neural model trained with generated and logged inputs and evaluated on PCRE… | ||
| MArk: Exploiting Cloud Services for Cost-Effective, SLO-Aware Machine Learning Inference Serving | Chengliang Zhang, Minchen Yu, Wei Wang, Feng Yan | 20m |
MArk is an AWS inference-serving system that combines dynamic batching (to make GPUs cheaper than CPUs at moderate batch sizes), predictive autoscaling (to… | ||
| Cross-dataset Time Series Anomaly Detection for Cloud Systems | Xu Zhang, Qingwei Lin, Yong Xu, Si Qin, Hongyu Zhang, Bo Qiao, Yingnong Dang, Xinsheng Yang, Qian Cheng, Murali Chintalapati, Youjiang Wu, Ken Hsieh, Kaixin Sui, Xin Meng, Yaohai Xu, Wenchi Zhang, Furao Shen, Dongmei Zhang | 21m |
ATAD trains anomaly detectors on labelled cloud datasets and applies them to new unlabelled services by combining transfer learning with active learning… | ||
| nothing matches | ||