logo资料库

OSDI2018 paper集合.pdf

发布时间：2022-06-23 发布人：admin 分类：说明书 资料大小：48.93M 资料格式：pdf 举报 版权申诉
第1页 / 共827页
第2页 / 共827页
第3页 / 共827页
第4页 / 共827页
第5页 / 共827页
第6页 / 共827页
第7页 / 共827页
第8页 / 共827页
资料共827页，剩余部分请下载后查看
    osdi18-huang
    Introduction
    Problem Statement
    Panorama System
    Overview
    Abstractions and APIs
    Local Observation Store
    Observers
    Observation Exchange
    Judging Failure from Observations
    Design Pattern and Observability
    A Failed Case
    Observability Patterns
    Implications
    Observability Analysis
    Locate Observation Boundary
    Identify Observer and Observed
    Extract Observation
    Handling Indirection
    Implementation
    Evaluation
    Experiment Setup
    Integration with Several Systems
    Detection of Crash Failures
    Detection of Gray Failures
    Fault Localization
    Transient Failure, Normal Operations
    Performance
    Discussion and Limitations
    Related Work
    Conclusion
    osdi18-cui
    Introduction
    Overview
    Problem Statement
    Design Choices
    Challenges
    Irreversible Instructions
    Missing Memory Writes
    Concurrent Memory Writes
    Design
    Instruction Reversal
    Irreversible Instruction Handling
    Recovering Memory Writes
    Data Inference Graph
    Error Correction
    Handling Concurrency
    Implementation
    Online Hardware Tracing
    Offline Binary Analysis
    Deployment
    Evaluation
    Accuracy
    Single-Thread Accuracy
    Multiple-Thread Accuracy
    Efficiency
    Effectiveness
    Deployment
    Discussion
    Related Work
    Conclusion
    Acknowledgments
    osdi18-mohan
    Introduction
    Background
    Studying Crash-Consistency Bugs
    B3: Bounded Black-Box Crash Testing
    Overview
    Bounds used by B3
    Fine-grained correctness checking
    Limitations
    CrashMonkey and Ace
    CrashMonkey
    Automatic Crash Explorer (Ace)
    Testing and Bug Analysis
    Evaluation
    Experimental Setup
    Bug Finding
    CrashMonkey Performance
    Ace Performance
    Resource Consumption
    Related Work
    Conclusion
    osdi18-alquraan
    osdi18-shan
    osdi18-cutler
    osdi18-khawaja
    osdi18-maeng
    osdi18-qin
    osdi18-yang
    osdi18-sriraman
    Introduction
    Motivation
    A Taxonomy of Threading Models
    Key dimensions
    Synchronous models
    Asynchronous models
    Tune: System Design
    Implementation
    Experimental Setup
    Evaluation
    Threading model characterization
    Synchronous vs. Asynchronous
    Synchronous models
    Asynchronous models
    Load adaptation
    Comparison to the state-of-the-art
    Steady-state adaptation
    Load transients
    Discussion
    Related Work
    Conclusion
    Acknowledgement
    References
    osdi18-berger
    Introduction
    Background and Challenges
    The goal of RobinHood
    Challenges of caching for tail latency
    Time-varying latency imbalance
    Latency is not correlated with specific queries nor with query rate
    Latency depends on request structure, which varies greatly
    The RobinHood Caching System
    The basic RobinHood algorithm
    Refining the RobinHood algorithm
    RobinHood architecture
    System Implementation and Challenges
    Implementation and testbed
    Implementation challenges
    Generating experimental data
    Evaluation
    Competing caching systems
    How much does RobinHood improve SLO violations for OneRF's workload?
    How much variability can RobinHood handle?
    How robust is RobinHood to simultaneous latency spikes?
    How much space does RobinHood save?
    What is the overhead of running RobinHood?
    Discussion
    Related Work
    Conclusion
    osdi18-gjengset
    osdi18-wei
    osdi18-mahajan
    osdi18-hsieh
    1 Introduction
    2 Background and Motivation
    2.1 Convolutional Neural Networks
    2.2 Characterizing Real-world Videos
    2.2.1 Excluding large portions of videos
    2.2.2 Limited set of object classes in each video
    2.2.3 Feature vectors for finding duplicate objects
    3 Overview of Focus
    4 Video Ingest & Querying Techniques
    4.1 Approximate Index via Cheap Ingest
    4.2 Video-specific Specialization of Ingest CNN
    4.3 Redundant Object Elimination
    4.4 Trading off Ingest Cost and Query Latency
    5 Implementation
    5.1 Ingest Processor
    5.2 Stream Tuner
    5.3 Query Processor
    6 Evaluation
    6.1 Methodology
    6.2 End-to-End Performance
    6.3 Effect of Different Focus Components
    6.4 Ingest Cost vs. Query Latency Trade-off
    6.5 Sensitivity to Recall/Precision Target
    6.6 Sensitivity to Object Class Numbers
    7 Other Applications
    8 Related Work
    9 Conclusion
    osdi18-sigurbjarnarson
    osdi18-chajed
    Introduction
    Related work
    Goal and challenges
    Approach to proving atomicity
    Design of Cspec
    Layers
    Defining correctness
    Implementation
    Proving
    Cspec's proof patterns
    Mover pattern
    Protocol pattern
    Abstraction pattern
    Other patterns
    Implementation
    Evaluation
    Verifying Cmail
    Verifying a counter on x86-TSO
    Speedup
    Effort
    Patterns
    Trusted computing base
    Conclusion
    osdi18-ileri
    Introduction
    Related Work
    Motivation: bugs
    Goal
    Threat model
    Challenges
    Specification: data noninterference
    Proof approach: sealed blocks
    Formalizing sealed blocks
    Proving noninterference
    Code generation
    Case study: File system
    Specifying security
    Modifying the implementation
    Proving security
    Implementation
    Evaluation
    Specification trustworthiness
    Effort
    Performance
    Conclusion
    osdi18-setty
    1 Introduction
    2 Problem statement and background
    2.1 A prior instantiation of VSMs
    2.1.1 Interacting with external resources
    2.1.2 Handling state
    2.2 Outlook and roadmap
    3 Efficient storage operations in VSMs
    3.1 SetKV: A verifiable key-value store
    3.2 Building VSMs using SetKV
    4 Supporting concurrent services
    4.1 Executing requests concurrently
    4.2 Supporting transactional semantics
    5 Efficient instantiations
    5.1 Parallelizing audits
    5.2 Efficient cryptographic primitives
    6 Implementation and applications
    6.1 Applications of Spice
    7 Experimental evaluation
    7.1 Spice's approach to state VS. prior solutions
    7.2 Benefits of Spice's concurrent execution
    7.3 Performance of apps built with Spice
    8 Related work
    9 Discussion and summary
    osdi18-lockerman
    osdi18-veeraraghavan
    osdi18-alagappan
    osdi18-maricq
    osdi18-klimovic
    Introduction
    Storage for Serverless Analytics
    Ephemeral Storage Requirements
    Existing Systems
    Pocket Design
    System Architecture
    Application Interface
    Life of a Pocket Application
    Handling Node Failures
    Rightsizing Resource Allocations
    Rightsizing Application Allocation
    Rightsizing the Storage Cluster
    Implementation
    Evaluation
    Methodology
    Microbenchmarks
    Rightsizing Resource Allocations
    Comparison to S3 and Redis
    Discussion
    Related Work
    Conclusion
    osdi18-annamalai
    Introduction
    Motivation
    Capital and operational costs matter
    Service request movements
    Low read-write ratios
    Ineffectiveness of distributed caches
    Separate locality management layer
    Background
    Akkio Design and Implementation
    Design guidelines
    Requirements
    Architectural overview
    Akkio Location Service (ALS)
    Access Counter Service
    Akkio Data Placement Service (DPS)
    Provisioning new µ-shards
    Determining optimal µ-shard placement
    µ-shard migration
    Replica set collection changes
    DPS fault recovery
    Evaluation
    Implementation metrics
    Use cases analysis
    ViewState
    AccessState
    Instagram Connection-Info
    Instagram Direct
    Analysis of Akkio services
    Related Work
    Concluding Remarks
    osdi18-zuo
    Introduction
    Background and Motivation
    Data Consistency in NVM
    Hashing Index Structures for NVM
    Conventional Hashing Schemes
    Hashing Schemes for NVM
    Resizing a Hash Table
    The Level Hashing Design
    Write-optimized Hash Table Structure
    Cost-efficient In-place Resizing
    Low-overhead Consistency Guarantee
    Concurrent Level Hashing
    Performance Evaluation
    Experimental Setup
    Experimental Results
    Maximum Load Factor
    Insertion Latency
    Update Latency
    Search Latency
    Deletion Latency
    Resizing Time
    Concurrent Throughput
    Related Work
    Conclusion
    osdi18-zhang
    osdi18-bhagwan
    osdi18-jindal
    Introduction
    Key Insights
    Competing/similar apps are abundant
    Similar apps differ in energy drain
    Framework services dominate app energy drain
    How to Diff Energy Profiles?
    What diffing granularity?
    How do app tasks manifest in call trees?
    What tree structures to diff?
    How to perform Eflask matching?
    Need for approximate DiffProf matching
    Prior tree matching algorithms
    The Eflask matching algorithm
    Preprocessing CCTs to facilitate effective matching
    Implementation and Usage
    Evaluation
    Experimental setup
    Diffing results
    Effectiveness
    Unmatched (extra) tasks
    Matched tasks
    Discussions
    Related work
    Conclusion
    osdi18-zhou
    osdi18-quinn
    Introduction
    Usage
    Debugging tools
    Parallel retro-logging
    Continuous function evaluation
    Retro-timing
    Scenarios
    Atomicity Violation
    Apache 45605
    Apache 25520
    Data corruption
    Wild store
    Memory leak
    Lock Contention
    Design and implementation
    Background: Deterministic record and replay
    Sledgehammer API
    Tracers
    Tracepoints
    Preparing for debugging queries
    Running a parallel debugging tool
    Isolation
    Support for continuous function evaluation
    Support for retro-timing
    Aggregating results
    Evaluation
    Experimental Setup
    Benchmarks
    Scalability
    Scaling bottlenecks
    Benefit of parallel analysis
    Isolation
    Recording Overhead
    Related Work
    Conclusion
    osdi18-moritz
    osdi18-chen
    Introduction
    Overview
    Optimizing Computational Graphs
    Generating Tensor Operations
    Tensor Expression and Schedule Space
    Nested Parallelism with Cooperation
    Tensorization
    Explicit Memory Latency Hiding
    Automating Optimization
    Schedule Space Specification
    ML-Based Cost Model
    Schedule Exploration
    Distributed Device Pool and RPC
    Evaluation
    Server-Class GPU Evaluation
    Embedded CPU Evaluation
    Embedded GPU Evaluation
    FPGA Accelerator Evaluation
    Related Work
    Conclusion
    osdi18-xiao
    osdi18-lee
    Introduction
    Model Serving: State-of-the-Art and Limitations
    White Box Prediction Serving: Design Principles
    The Pretzel System
    Off-line Phase
    Flour
    Oven
    Object Store
    On-line Phase
    Runtime
    Scheduler
    Additional Optimizations
    Evaluation
    Memory
    Latency
    Micro-benchmark
    End-to-end
    Throughput
    Heavy Load
    Micro-benchmark
    End-to-end
    Limitations and Future Work
    Related Work
    Conclusion
    osdi18-kulkarni
    osdi18-yeo
    osdi18-phothilimthana
    osdi18-volos
    osdi18-konoth
    Introduction
    Background
    DRAM Organization
    The Rowhammer Bug
    Rowhammer Defenses
    Threat Model
    Design
    Implementation
    ZebRAM Prototype Components
    Implementation Details
    Security Evaluation
    Traditional Rowhammer Exploits
    ZebRAM-aware Exploits
    Attacking the Unsafe Region
    Attacking the Safe Region
    Performance Evaluation
    Related work
    Discussion
    Prototype
    Alternative Implementations
    Conclusion
    osdi18-lazar
    osdi18-crooks
    osdi18-iyer
    osdi18-wang
    Introduction
    Problems with State-of-the-Art Systems
    Challenges and Contributions
    Summary of Results
    Background and Overview
    Background
    RStream Overview
    Programming Model
    RStream Implementation
    Preprocessing
    Join Implementation
    Redundancy Removal via Automorphism Checks
    Pattern Aggregation via Isomorphism Checks
    Evaluation
    Comparisons with Mining Systems
    Comparisons with Datalog Engines
    RStream Performance Breakdown
    Related Work
    Conclusion
    osdi18-kalavri
    osdi18-essertel
    Blank Page
    i P r o c e e d n g s o f t h e 1 3 t h U S E N X S y m p o s u m o n O p e r a t i I i n g S y s t e m s D e s g n a n d I i l m p e m e n t a t i o n C a r l s b a d , C A , U S A O c t o b e r 8 – 1 0 , 2 0 1 8 ISBN 978-1-931971-47-8 conference proceedings 13th USENIX Symposium on Operating Systems Design and Implementation Carlsbad, CA, USA October 8–10, 2018 Sponsored by In cooperation with ACM SIGOPS
    OSDI ’18 Sponsors Gold Sponsors Silver Sponsors Bronze Sponsors Industry Partners and Media Sponsors ACM Queue FreeBSD Foundation © 2018 by The USENIX Association All Rights Reserved This volume is published as a collective work. Rights to individual papers remain with the author or the author’s employer. Permission is granted for the noncommercial reproduction of the complete work for educational or research purposes. Permission is granted to print, primarily for one person’s exclusive use, a single copy of these Proceedings. USENIX acknowledges all trademarks herein. ISBN 978-1-931971-47-8 USENIX Supporters USENIX Patrons Facebook • Google • Microsoft NetApp • Private Internet Access USENIX Benefactors Amazon • Bloomberg • Oracle Squarespace • VMware USENIX Partners Booking.com • Can Stock Photo • Cisco Meraki Fotosearch • Teradactyl • TheBestVPN.com Open Access Publishing Partner PeerJ
    USENIX Association Proceedings of the 13th USENIX Symposium on Operating Systems Design and Implementation October 8–10, 2018 Carlsbad, CA, USA
    Symposium Organizers Program Co-Chair Andrea Arpaci-Dusseau, University of Wisconsin— Madison Geoff Voelker, University of California, San Diego Program Committee Rachit Agarwal, Cornell University Marcos K. Aguilera, VMware Research Mahesh Balakrishnan, Yale University/Facebook Ranjita Bhagwan, Microsoft Research India Edouard Bugnion, École Polytechnique Fédérale de Lausanne (EPFL) Miguel Castro, Microsoft Research Cambridge Kang Chen, Tsinghua University Vijay Chidambaram, The University of Texas at Austin Landon Cox, Duke University Angela Demke Brown, University of Toronto Sandhya Dwarkadas, University of Rochester Sasha Fedorova, University of British Columbia Roxana Geambasu, Columbia University Cristiano Giuffrida, Vrije Universiteit Amsterdam Haryadi Gunawi, University of Chicago Andreas Haeberlen, University of Pennsylvania Jon Howell, VMware Research Rebecca Isaacs, Twitter Michael Isard, Google Frans Kaashoek, Massachusetts Institute of Technology Manos Kapritsos, University of Michigan Taesoo Kim, Georgia Institute of Technology Sam King, University of California, Davis Christoforos Kozyrakis, Stanford University Jinyang Li, New York University Wyatt Lloyd, Princeton University Jay Lorch, Microsoft Research Richard Mortier, University of Cambridge Gilles Muller, Inria KyoungSoo Park, Korea Advanced Institute of Science and Technology (KAIST) Raluca Popa, University of California, Berkeley Don Porter, The University of North Carolina at Chapel Hill Justine Sherry, Carnegie Mellon University Liuba Shrira, Brandeis University Ryan Stutsman, University of Utah Steve Swanson, University of California, San Diego Michael Swift, University of Wisconsin—Madison Dan Tsafrir, VMware Research and Technion—Israel Institute of Technology Rashmi Vinayak, Carnegie Mellon University Xi Wang, University of Washington Andrew Warfield, Amazon Roger Wattenhofer, ETH Zurich Hakim Weatherspoon, Cornell University Ming Wu, Microsoft Research Yubin Xia, Shanghai Jiao Tong University Ding Yuan, University of Toronto Matei Zaharia, Stanford University Irene Zhang, Microsoft Research Yiying Zhang, Purdue University Poster Session Co-Chairs Vijay Chidambaram, The University of Texas at Austin Yiying Zhang, Purdue University Steering Committee Brad Chen, Google Jason Flinn, University of Michigan Casey Henderson, USENIX Association Kimberly Keeton, Hewlett Packard Labs Hank Levy, University of Washington James Mickens, Harvard University Brian Noble, University of Michigan Timothy Roscoe, ETH Zurich Margo Seltzer, University of British Columbia Amin Vahdat, Google and University of California, San Diego
    Remzi Arpaci-Dusseau Anish Athalye Jonathan Behrens Gino Brunner Tej Chajed Nishanth Chandran Haibo Chen Charlie Curtsinger Cody Cutler Manuel Eichelberger Jon Gjengset Michael Gleicher Joseph Gonzalez Matthew Hicks Chris Hodsdon Atalay Mert İleri External Reviewers Pankaj Khanchandani David Lion Haonan Lu Yu Luo Tony Mason Darya Melnyk Ellis Michael Robert Morris Mihir Nanavati Luke Nelson Khiem Ngo Amy Ousterhout Pál András Papp Ali Razeen Xiang Ren Oliver Richter Kirk Rodrigues Edo Roth Brian Sandler Malte Schwarzkopf Igor Smolyar Zhenyu Song Theano Stavrinos Julian Steger Simon Tanner Yuyi Wang Michael Wei Tian Yang Idan Yaniv Yongle Zhang Xu Zhao Aviad Zuck
    Message from the OSDI ’18 Program Co-Chairs Dear colleagues, Welcome to the 13th USENIX Symposium on Operating Systems Design and Implementation (OSDI ’18), held in Carlsbad, CA, USA! This year’s technical program matches OSDI ’16 in including 47 exceptionally strong papers; these papers represent the many strengths of our community and cover a wide range of topics, including file and storage systems, networking, scheduling, security, formal verification of systems, graph processing, system support for machine learning, programming languages, fault-tolerance and reliability, debugging, and, of course, operating systems design and implementation. OSDI ’18 received 257 paper submissions, which the program committee reviewed in multiple rounds. Our program committee consisted of 49 reviewers with a mixture of academic and industrial research and practical experience. The PC was divided into 24 “light” and 25 “heavy” members. All papers received three reviews in the first round; based on those reviews, 122 papers were selected to proceed to the second round. Second round papers received a minimum of two additional reviews from heavy PC members. For a small number of papers, where opinions were divided or where a paper was particularly specialized, we solicited additional expert reviews. In total, the PC and external reviewers wrote over 2^10 reviews. As in previous OSDI review cycles, this year’s process included a response period in which authors could answer reviewer questions and address factual errors in the initial reviews. Authors of 191 papers submitted a response. Responses had a measurable impact on both our online and in-person discussions, and the author responses and ensuing online discussion influenced some PC members to adjust their reviews and reconsider their ratings. Overall, we believe author responses helped improve the quality of the selected program. After more than a week of online discussion across the full PC, we picked 83 papers for the heavy PC members to discuss at a 1.5-day PC meeting held at the University of Wisconsin in Madison, WI, USA. Almost all heavy PC members were able to attend in person, with just one person calling in remotely. As PC chairs, we strove to ensure that all the discussed papers received full and fair consideration, coming to a consensus agreement in most cases. Papers were placed into high-level categories according to their main topic so that similar papers could be discussed together at the PC meeting. All discussed papers received a summary of the PC discussion written by a heavy PC member. In the end, the PC selected 47 papers for presentation at the conference, resulting in an 18% acceptance rate. Each of the accepted papers was allocated an additional two pages and shepherded by a member of the heavy PC to help the authors address the reviewers’ comments in their camera-ready versions. After finalizing the program, we created a separate committee to decide the Jay Lepreau Best Paper Awards com- posed of PC members with no conflicts with the papers under consideration. PC members could nominate papers for these awards in their reviews or directly to us. We selected six papers with at least two votes for best paper as can- didates for the award. After reading the nominated papers and considering the reviews from the full PC, the awards committee agreed on three Jay Lepreau Best Paper Awards. As PC co-chairs, we stand on the shoulders of so many who did a tremendous amount of hard work to make OSDI ’18 a success. First, we thank the authors of all submitted papers for choosing to send their work to OSDI. Thanks also to the program committee for their hard work in reviewing and discussing the submissions and in shepherding the accepted papers. We particularly thank Yiying Zhang and Vijay Chidambaram for organizing an extensive poster session of more than 83 posters to be presented across two evenings. We are also grateful to the external reviewers who provided additional perspectives. We thank the USENIX staff, who have been fundamental in organizing OSDI ’18. Finally, OSDI wouldn’t be what it is without our attendees—thank you for listening to our speakers, asking challenging and insightful questions, sharing your ideas with others, and networking with one another in the hallways! We hope you will find OSDI ’18 interesting, educational, and inspiring! Andrea Arpaci-Dusseau, University of Wisconsin-Madison Geoff Voelker, University of California, San Diego OSDI ’18 Program Co-Chairs
    OSDI ’18: 13th USENIX Symposium on Operating Systems Design and Implementation October 8–10, 2018 Carlsbad, CA, USA Understanding Failures Capturing and Enhancing In Situ System Observability for Failure Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Peng Huang, Johns Hopkins University; Chuanxiong Guo, ByteDance Inc.; Jacob R. Lorch and Lidong Zhou, Microsoft Research; Yingnong Dang, Microsoft REPT: Reverse Debugging of Failures in Deployed Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Weidong Cui and Xinyang Ge, Microsoft Research Redmond; Baris Kasikci, University of Michigan; Ben Niu, Microsoft Research Redmond; Upamanyu Sharma, University of Michigan; Ruoyu Wang, Arizona State University; Insu Yun, Georgia Institute of Technology Finding Crash-Consistency Bugs with Bounded Black-Box Crash Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 Jayashree Mohan, Ashlie Martinez, Soujanya Ponnapalli, and Pandian Raju, University of Texas at Austin; Vijay Chidambaram, University of Texas at Austin and VMware Research An Analysis of Network-Partitioning Failures in Cloud Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Ahmed Alquraan, Hatem Takruri, Mohammed Alfatafta, and Samer Al-Kiswany, University of Waterloo Operating Systems LegoOS: A Disseminated, Distributed OS for Hardware Resource Disaggregation . . . . . . . . . . . . . . . . . . . . . . 69 Yizhou Shan, Yutong Huang, Yilun Chen, and Yiying Zhang, Purdue University The benefits and costs of writing a POSIX kernel in a high-level language . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 Cody Cutler, M. Frans Kaashoek, and Robert T. Morris, MIT CSAIL Sharing, Protection, and Compatibility for Reconfigurable Fabric with AmorphoS . . . . . . . . . . . . . . . . . . . . 107 Ahmed Khawaja, Joshua Landgraf, and Rohith Prakash, UT Austin; Michael Wei and Eric Schkufza, VMware Research Group; Christopher J. Rossbach, UT Austin and VMware Research Group Adaptive Dynamic Checkpointing for Safe Efficient Intermittent Computing . . . . . . . . . . . . . . . . . . . . . . . . . 129 Kiwan Maeng and Brandon Lucia, Carnegie Mellon University Scheduling Arachne: Core-Aware Thread Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 Henry Qin, Qian Li, Jacqueline Speiser, Peter Kraft, and John Ousterhout, Stanford University Principled Schedulability Analysis for Distributed Storage Systems using Thread Architecture Models . . . 161 Suli Yang, Ant Financial Services Group; Jing Liu, Andrea C. Arpaci-Dusseau, and Remzi H. Arpaci-Dusseau, UW-Madison µTune: Auto-Tuned Threading for OLDI Microservices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177 Akshitha Sriraman and Thomas F. Wenisch, University of Michigan RobinHood: Tail Latency Aware Caching – Dynamic Reallocation from Cache-Rich to Cache-Poor . . . . . . 195 Daniel S. Berger and Benjamin Berg, Carnegie Mellon University; Timothy Zhu, Pennsylvania State University; Siddhartha Sen, Microsoft Research; Mor Harchol-Balter, Carnegie Mellon University (continued on next page)
    分享到：
    赞 收藏

    相关推荐

    资料库

    操作系统

    共收录1433份资料，累计11个分类，关注成员有19位，主要包括：CentOS,MacOS,DOS,Solaris,Ubuntu,Unix,OS,桌面系统,其它,Windows Server,Linux

    热门标签

    CentOS MacOS DOS Solaris Ubuntu Unix OS 桌面系统 其它 Windows Server Linux

    最新资料