sosp2011 socc2011 plos2011 Report 23rd ACM Symposium on Operating Systems Principles (SOSP)October 23-26, 2011, Cascais, Portugal http://sosp2011.gsd.inesc-id.pt/ 2nd ACM Symposium on Cloud Computing October 26-28, 2011, Cascais, Portugal http://socc2011.gsd.inesc-id.pt/ 6th Workshop on Programming Languages and Operating Systems, October 23, 2011, Cascais, Portugal http://plosworkshop.org/2011/

1. SOSP’11 & SOCC’11 & PLOS’11
Report
at 29回仮想化実装技術勉強会(vimpl)
2011/Nov/21
須崎有康

2. SOSP2011概要
• 23rd ACM Symposium on Operating Systems Principles (SOSP)
October 23-26, 2011, Cascais, Portugal
– http://sosp2011.gsd.inesc-id.pt/
– 論文＆スライド＆ビデオ http://sigops.org/sosp/sosp11/current/index.html
– 1colum(screen)と2colums(print)の論文があり。便利だが、screen版が作りが甘い
– 170投稿、26採択。
– 参加者540(今まで最高。そのうちStudentsが220)
– Best Paper 2本
• ColumbiaのCells: A Virtual Mobile Smartphone Architecture
• Wisconsin, MadisonのA File is Not a File: Understanding the I/O Behavior
of Apple Desktop Applications
– ポスター30(投稿50) 。日本から1件(産総研) (最終版投稿にshepherdが付く)
– WIP 17(投稿48)
– 次回は中国？
• ワークショップ
– SLAML - Managing Large-Scale Systems via the Analysis of System Logs and
the Application of Machine Learning Techniques
– PLOS - 6th Workshop on Programming Languages and Operating Systems
– MobiHeld - 3rd ACM SOSP Workshop on Networking, Systems, and
Applications on Mobile Handhelds
– HotPower - 4th Workshop on Power-Aware Computing and Systems

3. プログラム１日目 午前 1/2
• Key-Value Chair: Marvin Theimer
– SILT: A Memory-Efficient, High-Performance
Key-Value Store
• Hyeontaek Lim, Bin Fan, David G. Andersen (CMU), Michael Kaminsky (Intel Labs)
• 小メモリを特徴とするKVS。 ソースコード https://github.com/silt/silt/
• Memory Efficiency (FAWN[OSDI09], FlashStore[VLDB10], BUfferHash[NDSI10],
HashCache[NSDI09]) VS High performance (SkimpyStash[SIGMOD11])
– Scalable Consistency in Scatter
• Lisa Glendenning, Ivan Beschastnikh, Arvind Krishnamurthy, Thomas Anderson (University of Washington)
• CcalabilityとConsistencyを両立するScatterの提案。Consistencyの使い方を突っ込まれていた。
– Fast Crash Recovery in RAMCloud
• Diego Ongaro, Stephen M. Rumble, Ryan Stutsman, John Ousterhout, Mendel Rosenblum (Stanford)
• RAMの問題はPower Failure。ディスクからの高速リカバリが課題。３つのディスクコピーをScatter log
Data（各8MB） として保存。これからPartitioned Recovery する。

4. プログラム１日目 午前 2/2
• Storage Chair: Eddie Kohler
– Design Implications for Enterprise Storage Systems via Multi-Dimensional
Trace Analysis
• Yanpei Chen (UC Berkeley), Kiran Srinivasan, Garth Goodson (NetApp), Randy Katz (UC
Berkeley)
• 16種類のdimension (Total IO size , Read/Write, request, average time between IO, read/write
sequence, etc)で解析。
• 70％以上がシーケンシャルread/wireアクセス。 I/Oのタイプで.xls, .ppt, .doc, .dpfなどのファイルが推定
できる。
• NTTでもiSCSI経由で振る舞い解析からファイルタイプを推定する特許があるらしい。
– Differentiated Storage Services
g
• Michael Mesnier, Jason B. Akers, Feng Chen (Intel), Tian Luo (Ohio State)
• ディスクはSSDにキャッシュされるようになったが、大きなシーケンシャルファイルアクセスはキャッシュを
バイパスさせる。ファイルシステムにこの機能を持たせる。
– （Best Paper）A File is Not a File: Understanding the I/O Behavior of Apple
Desktop Applications (詳細後述)
• Tyler Harter, Chris Dragga, Michael Vaughn, Andrea C. Arpaci-Dusseau, Remzi H. Arpaci-Dusseau (University of
Wisconsin, Madison)

5. プログラム１日目 午後
• Security Chair: Adrian Perrig
– CryptDB: Protecting Confidentiality with Encrypted Query Processing
• Raluca Ada Popa, Catherine M. S. Redfield, Nickolai Zeldovich, Hari Balakrishnan (MIT)
– Intrusion Recovery for Database-backed Web Applications
• Ramesh Chandra, Taesoo Kim, Meelap Shah, Neha Narula, Nickolai Zeldovich (MIT)
– Software fault isolation with API integrity and multi-principal modules
• Yandong Mao, Haogang Chen (MIT), Dong Zhou (Tsinghua), Xi Wang, Nickolai Zeldovich, M. Frans Kaashoek (MIT)
• カーネルモジュールを守るLXFIの提案。XFI[OSDI06]、BGI[SOSP09]からinspire。 APIは悪用される可
能性がある。Annotation langで記述 LXFIがコンパイル時にrun time コードを埋め込む。
• Reality Chair: George Candea
y g
– Thialfi: A Client Notification Service for Internet-Scale Applications
• Atul Adya, Gregory Cooper, Daniel Myers, Michael Piatek (Google)
– Windows Azure Storage: A Highly Available Cloud Storage Service with
Strong Consistency
• Brad Calder, Ju Wang, Aaron Ogus, Niranjan Nilakantan, Arild Skjolsvold, Sam McKelvie, Yikang Xu, Shashwat Srivastav, Jiesheng
Wu, Huseyin Simitci, Jaidev Haridas, Chakravarthy Uddaraju, Hemal Khatri, Andrew Edwards, Vaman Bedekar, Shane Mainali,
Rafay Abbasi, Arpit Agarwal, Mian Fahim ul Haq, Muhammad Ikram ul Haq, Deepali Bhardwaj, Sowmya Dayanand, Anitha
Adusumilli, Marvin McNett, Sriram Sankaran, Kavitha Manivannan, Leonidas Rigas (Microsoft)
– An Empirical Study on Configuration Errors in Commercial and Open
Source Systems
• Zuoning Yin, Xiao Ma (UIUC), Jing Zheng, Yuanyuan Zhou (UCSD), Lakshmi N. Bairavasundaram, Shankar
Pasupathy (NetApp)
• Posters (発表詳細後述)

6. プログラム２日目 午前 1/2
• Virtualization Chair: Gernot Heiser
– (Best Paper) Cells: A Virtual Mobile Smartphone Architecture
• Jeremy Andrus, Christoffer Dall, Alex Van’t Hof, Oren Laadan, Jason Nieh (Columbia)
• Smart Phoneの仮想化。オーバーヘッド2%。ベンチャー創業 http://www.cellrox.com/
– Breaking Up is Hard to Do: Security and Functionality in a Commodity
Hypervisor (詳細後述)
• Patrick Colp, Mihir Nanavati (UBC), Jun Zhu (Citrix), William Aiello (UBC), George Coker (NSA), Tim Deegan
(Citrix), Peter Loscocco (NSA), Andrew Warfield (UBC)
– CloudVisor: Retrofitting Protection of Virtual Machines in Multi-tenant
Cloud with Nested Virtualization (詳細後述)
• Fengzhe Zhang, Jin Chen, Haibo Chen, Binyu Zang (Fudan University)
– (Audience Choice) Atlantis: Robust, Extensible Execution Environments for
Web Applications
• James Mickens (MSR), Mohan Dhawan (Rutgers)
• 関連研究。 Illinois Browser Operating System (IBOS) [OSDI10], OP[SS&`P08]

7. プログラム２日目 午前 2/2
• OS Architecture Chair: Nickolai Zeldovich
– PTask: Operating System Abstractions To Manage GPUs as Compute
Devices
• Christopher J. Rossbach, Jon Currey (MSR), Mark Silberstein (Technion), Baishakhi Ray, Emmett Witchel (UT
Austin)
• CPUのロードの割合がGPUの性能に影響する。逆も同様。
• GPUスケジューリングを抽象化するPtask (Parallel Task)。 data flowを抽象化。Fairnessのために
priorityあり。通常のフィルタとインターフェースとできる。
• 例：#> capture(process) | xform(ptask) | filer (ptask)
– Logical Attestation: An Authorization Architecture for Trustworthy
Computing
C ti
• Emin Gün Sirer (Cornell), Willem de Bruijn (Google), Patrick Reynolds (BlueStripe Software), Alan Shieh, Kevin
Walsh, Dan Williams, Fred B. Schneider (Cornell)
• Trusted Computingの話。
• Cornellで作っているlogical attestationに対応するOSはNexusと言う。 NexusのMicro kernel 自体に
Introspection の機能がある。 NexusにはFauxbook と言うprivacy-preserving social networkが入って
いる。

8. プログラム２日目 午後
• Detection and Tracing Chair: Rebecca Isaacs
– Practical Software Model Checking via Dynamic Interface Reduction
• Huayang Guo (MSR and Tsinghua), Ming Wu, Lidong Zhou (MSR), Gang Hu (MSR and Tsinghua), Junfeng Yang
(Columbia), Lintao Zhang (MSR)
– Detecting failures in distributed systems with the FALCON spy network
• Joshua B. Leners, Hao Wu, Wei-Lun Hung (UT Austin), Marcos K. Aguilera (MSR), Michael Walfish (UT Austin)
– Secure Network Provenance
• Wenchao Zhou, Qiong Fei, Arjun Narayan, Andreas Haeberlen, Boon Thau Loo (University of Pennsylvania), Micah
Sherr (Georgetown University)
– Fay: Extensible Distributed Tracing from Kernels to Clusters
• Úlfar Erlingsson (G
Úlf E li (Google), M
l ) Marcus P i d (MSR) Si
Peinado (MSR), Simon P
Peter (ETH Zürich), Mih i B di (MSR)
Zü i h) Mihai Budiu
• Work in Progress
– 48 submissions, 17 accepted. 一人４分。
– ASPLOS11 で発表されたDoublePlayの拡張
– Paxosに対抗するIsis^2 http://www.cs.cornell.edu/ken/
– PROobE (Parallel Reconfigurable Observational Environment) http://newmexicoconsortium.org/probe

9. プログラム３日目 午前
• Threads and Races Chair: Bryan Ford
– Dthreads: Efficient Deterministic Multithreading
• Tongping Liu, Charlie Curtsinger, Emery D. Berger (UMass Amherst)
– Efficient Deterministic Multithreading through Schedule Relaxation
• Heming Cui, Jingyue Wu, John Gallagher, Huayang Guo, Junfeng Yang (Columbia)
– Pervasive Detection of Process Races in Deployed Systems
• Oren Laadan, Nicolas Viennot, Chia-che Tsai, Chris Blinn, Junfeng Yang, Jason Nieh (Columbia)
– Detecting and Surviving Data Races using Complementary Schedules
• Kaushik Veeraraghavan, Peter M. Chen, Jason Flinn, Satish Narayanasamy (University of Michigan)
• ASPLOS11でDoublePlay を発表した人
で y
• Geo-Replication Chair: Ant Rowstron
– Transactional storage for geo-replicated systems
• Yair Sovran, Russell Power (NYU), Marcos K. Aguilera (MSR), Jinyang Li (NYU)
– Don’t Settle for Eventual: Scalable Causal Consistency for Wide-Area
Storage with COPS
• Wyatt Lloyd, Michael J. Freedman (Princeton), Michael Kaminsky (Intel Labs), David G. Andersen (CMU)

10. A File is Not a File: Understanding the I/O Behavior of Apple
Desktop Applications (Best Paper)
Tyler Harter, Chris Dragga, Michael Vaughn, Andrea C. Arpaci-Dusseau, Remzi H. Arpaci-
Dusseau (University of Wisconsin, Madison)
• ホームユーザアプリケーション(MaxOS X対象)は、UNIX流の
単純なアプリケーションとは異なり、大規模で複雑。解析するた
めにiBenchを提案。
– http://research.cs.wisc.edu/adsl/Traces/ibench/
• Case Study
– Pages 4.0.3(Apple s iWork suite)によるDOCフォーマット文書ファイル
4 0 3(Apple’s
作成。Insert 15 JPEG images (each ~2.5MB)
• 1docファイルの為に385 files
– 218 KV store files + 2 SQLite files: Personalized behavior (recently used lists, settings, etc)
– 118 multimedia files: Rich graphical experience
– 25 Strings files: Language localization
– 17 Other files: Auto-save file and others
• docファイル中にはディレクトリをもつ。(Document, Data, Table, Ministream)。
• 11 threads で実行
• fsync とrenameを多用した更新
– シーケンシャルアクセスしたつもりでもシーケンシャルになっていない
• FATではヘッダを更新した後に、各データストリームがフラグメント化されて保存

11. Th
hreads
Files
s
fsyncと
rename
の多用 small I/O
次スライド big I/O
で拡大
fsync
rename

12. Writing the
DOC file
Dodファイル
への書き込み
read
write

13. A File is Not a File: Understanding the I/O Behavior of Apple
Desktop Applications (Best Paper)
Tyler Harter, Chris Dragga, Michael Vaughn, Andrea C. Arpaci-Dusseau, Remzi H. Arpaci-
Dusseau (University of Wisconsin, Madison)
• 何故アプリケーションが大きくなるのか。
– 現在は開発環境による開発が主流
– Cocoa を使うと 689 のファイルから112,047行が取り込まれる
– 多言語対応も自動的に行われるが、多くのファイルが自動的に含まれる
• ホームユーザアプリケーションを解析するためにiBenchを提案
– iPhoto, iTunes, iMovie,Pages, Numbers, and Keynoteを組み合わせ
たBenchmark Suite
– AppleScriptによるUIのプログラミング
– Dtraceにsystem callの収集
• iTunesはptraceを使ってしまっているので、gdbを使って回避。
– ファイルサイズごとに解析
• very small (< 4KB), small (< 64KB), medium (< 1MB), large (< 10MB), very large
(≥ 10MB).

14. CloudVisor: Retrofitting Protection of Virtual Machines in
Multi-tenant Cloud with Nested Virtualization
Fengzhe Zhang, Jin Chen, Haibo Chen, Binyu Zang (Fudan University)
• 現在のVMMは大規模でTCB(Trusted Computing
Base)と呼ぶには大きすぎる。
• VMMの脆弱性も多い
VMMの脆弱性も多い。
– 32 CVE on Xen, 35 CVE on VMware
• Cloud Operatorによる情報漏えいもある。
• Nested VirtualizationによりVMM(Xen)の下位に置き、
VMメモリイメージ、I/Oを暗号化し、VM間や管理OSな
どからの不正アクセスを禁止するCloudVisorの提案。

15. CloudVisor: Retrofitting Protection of Virtual Machines in
Multi-tenant Cloud with Nested Virtualization
Fengzhe Zhang, Jin Chen, Haibo Chen, Binyu Zang (Fudan University)
• CloudVisor 5.5K LOC。Intel CPUの機能をフルに使う
• VMM起動後に挿入
– Intel TXTを使ったlate launch。Xenの起動後に挿入できる。
• IOMMUを使ってDMA Attackを防ぐ
• EPT-v ReadOnlyによるVMMメモリの仮想化とメモリ保護
• VMexit の回数が増える
– VM read と VM writeをVMCSへのメモリアクセスとするパッチ
で回避

16. CloudVisor: Retrofitting Protection of Virtual Machines in
Multi-tenant Cloud with Nested Virtualization
Fengzhe Zhang, Jin Chen, Haibo Chen, Binyu Zang (Fudan University)
• I/Oの暗号による秘匿化とハ
ッシュによる完全性確保
• Merkle hash treeによる
管理
• 性能比較のために全機能を実行できるマシンがない。２台で実験
– Core2Quad マシン Intel TXT, TPM
– Dell R510 EPT,AES-NI命令, SR-IOV
• 実装の疑問点
– CloudVisor用のメモリの確保。TXT後にどこに確保するのか。
– 暗号/復号の開始時点。VMが動く前にはCloudVisorを起動する必要あり?

17. CloudVisor: Retrofitting Protection of Virtual Machines in
Multi-tenant Cloud with Nested Virtualization
Fengzhe Zhang, Jin Chen, Haibo Chen, Binyu Zang (Fudan University)
• 類似研究
– Nested Virtualization (Turtles, OSDI[2010])
• Turtlesは複数VMMをターゲットとしているが、CloudVisorは一つ
– VMメモリの暗号化。
• OverShadow[ASPLOS08], SP3[Vee08]
– VMレイヤの縮小 強化
VMレイヤの縮小、強化
• NoHype[ISCA10]: virtualization layerの除去
• NOVA[EuroSys10]: microkernel based VMM
• HyperSentry [CCS10]: SMIを使ったVMM完全性検証
• HyerSafe[SS&P10]:制御フローの完全性確保
– Non-Bypassable Memory Lockdown, Restricted Pointer Indexing
• HUKO[NDSS11]：信頼できないドライバ用に隔離されたメモリ提供
– EPTでCR3をカーネル用とドライバ用に切り替える

18. Breaking Up is Hard to Do: Security and Functionality in a
Commodity Hypervisor
Patrick Colp, Mihir Nanavati (UBC), Jun Zhu (Citrix), William Aiello (UBC), George Coker (NSA), Tim
Deegan (Citrix), Peter Loscocco (NSA), Andrew Warfield (UBC)
• 現在のVMM(Xen)は
– Monolithic Control VM を含めてTCB(Trusted Computing Base)が大きい
– Control VM はVMに対して多くのインターフェース
• 空間的、時間的にmodularity とisolationを高めたXoarの提案
Xoarの構成
既存のXen

19. – 空間的
• Isolation：必要な機能毎のモジュール化、必要な特権権限のみ付与。
• Configurable Sharing
• Auditing
– 時間的：攻撃されるターゲットを必要な時のみ使うようにする
• Disposable： 起動後不要なものは捨てる。PCIの初期化など。
• Timed Restarts： デバイスは必要な時のみ
• Stateless： VMのロールバック。攻撃があってもアプリケーションを攻撃前に戻す
– 4-25msecで戻せる。副作用がある場合は反映させる。
使われる技術 recovery box [USENIX Summer 92] Microreboot[OSDI04]

20. • 各モジュールの性質
privileged “R” is rebootable

21. Poster
• 30(投稿50本) 最終版投稿にshepherdが付く
• Title: Software Side Channel Attack on Memory Deduplication
– 3 Referees: Weak Accept, Accept, Reject
– EuroSec2011で発表できなかったメモリ重複除外へのサイドチャネル攻撃
Normal Attacked
VM1(victim) VM2(attacker) VM1(victim) VM2(attacker) Write Access
Guest
Pseudo
Memory
Attacker knows the
Real page contents exist
Physical on victim VM.
Memory
Same-content pages
S t t When
Wh a write access is issued to the d d li
i i i d h deduplicated
d
on VMs are merged on pages, the access time is delayed, because the
real physical memory page are re-created by Copy-On-Write .
attacker’s program memory on attacker’s VM
matching
Un-control by
target file
… attacker’s program Page
Cache
open();
Contents stored
•Challenge for the attacker …
posix_memalign();
on page cache
read(); Heap
•Alignment of matching data … Contents stored on
heap memory
Same contents
are deduplicated
gettimeofday(); on self memory
•Self-reflection (explained to the right) // wrie data to heap
gettimeofday();
access time is delayed with self
memory deduplication
•Runtime modification
(swap-out, ASLR, anonymous page, matching target
file with gziped
attacker’s program memory on attacker’s VM
Un-control by
… attacker’s program
preloading, self-modifying code) gzopen();
Contents stored
Page
Cache
… on page cache different
posix_memalign(); contents
gzread();
Contents stored Heap
…
on heap memory
gettimeofday();
// write data to heap Deduplicated with
memory on other VM
gettimeofday();

22. SOSP報告終わり

23. SOCC2011概要
• 2nd ACM Symposium on Cloud Computing October 26-28, 2011, Cascais,
Portugal
– http://socc2011.gsd.inesc-id.pt/
– 採択論文 Long 24/136 (17%) short 6/42 (14%)
– 参加者194(昨年 208)
– 初日にTutorial 2本
• Amazon Web Services
• The Windows Azure Cloud Platform
– Keynote Speaker by Christopher Olston, Bionica Human Computing
y p y p p g
• Programming and Debugging Large-Scale Data Processing Workflows
– Hadoop MapReduce 用データフロー言語Pig、ワークフローマネージャNOVA、およ
びプログラミング環境の解説
• 異なる会議だがほぼ同じスライドを使っている。
• http://www.sfbayacm.org/wp/wp-content/uploads/2011/03/olston-slides.pdf

24. プログラム１日目 午前
• Virtual Infrastructure Chair: Amr El Abbadi
– Paper of Distinction! Pesto: Online Storage Performance
Management in Virtualized Datacenters [full paper]
• Ajay Gulati, Ganesha Shanmuganathan, and Irfan Ahmad (VMware, Inc.), Carl A.
Waldspurger (unaffiliated), and Mustafa Uysal (VMware, Inc.)
• データセンターのストレージパフォーマンスすシステム。
• PestoはCongestion management はPARDA[FAST09]ベース。I/O load
balancingによりデータの移動。Datastoreのremove/addあり。
– Improving Per-Node Efficiency in the Datacenter with New OS Abstractions [short
paper]
• Barret Rhoden, Kevin Klues, David Zhu, and Eric Brewer (UC Berkeley)
– Paper of Distinction! Opportunistic Flooding to Improve TCP Transmit
Performance in Virtualized Clouds [full paper]
• Sahan Gamage, Ardalan Kangarlou, Ramana Rao Kompella, and Dongyan Xu (Purdue University)
VMのコンテキストスイッチでRTTが悪くなる。ドライ
バドメインでOffload congestion controlするvFlood
を提案。類似研究 vSnoop [SC10]
packet lossがあった時はcongestion controlをVM
側に移す。
vFloodはXenのドライバドメインで実現。1500LOC。
Xen3.3 Linux.2.6.18 for driver domain
– Cuanta: Quantifying Effects of Shared On-chip Resource Interference for
Consolidated Virtual Machines [full paper]
• Sriram Govindan (The Pennsylvania State University), Jie Liu and Aman Kansal (Microsoft Research, Redmond), and
Anand Sivasubramaniam (The Pennsylvania State University)

25. プログラム１日目 午後 1/2
• Computation Chair: John Wilkes
– Orleans: Cloud Computing for Everyone [full paper]
• Sergey Bykov, Alan Geller, Gabriel Kliot, James R. Larus, Ravi Pandya, and Jorgen Thelin (Microsoft Research)
– Paper of Distinction! PrIter: A Distributed Framework for Prioritized Iterative
Computations [full paper]
• Yanfeng Zhang and Qixin Gao (Northeastern University, 中国), Lixin Gao (University of Massachusetts Amherst), and
Cuirong Wang (Northeastern University)
• MacPreduceの高速化。基本はIterative Computationのプライオリティを付けて、キューに
入れる。スレッシュホールドより大きいものから分散処理。Hadoopより50倍のスピードアップ。
• ソースコード http://code.google.com/p/priter/
– Making Time-stepped Applications Tick in the Cloud [full paper, regular
presentation]
• Tao Zou, Guozhang Wang, Marcos Vaz Salles, David Bindel, Alan Demers, Johannes Gehrke, and Walker White (Cornell
University)
– Scaling the Mobile Millennium System in the Cloud [short paper]
• Timothy Hunter, Teodor Moldovan, Matei Zaharia, and Justin Ma (UC Berkeley), Samy Merzgui (EPFL), and Justin Ma,
Michael Franklin, Pieter Abbeel, and Alexandre Bayen (UC Berkeley)

26. プログラム１日目 午後 2/2
• Data Centers etc Chair: Jeff Chase
– ALIAS: Scalable, Decentralized Label Assignment for Data Centers [full paper]
• Meg Walraed-Sullivan, Radhika Niranjan Mysore, and Malveeka Tewari (UCSD), Ying Zhang (Ericsson), and Keith Marzullo
and Amin Vahdat (UCSD)
– Small-World Datacenters [full paper]
• Ji-Yong Shin, Bernard Wong, and Emin Gun Sirer (Cornell University)
– Switching the Optical Divide: Fundamental Challenges for Hybrid Electrical/Optical
Datacenter Networks [short paper]
• Hamid Hajabdolali Bazzaz and Malveeka Tewari (University of California, San Diego), Guohui Wang (Rice University),
George Porter (University of California, San Diego), T. S. Eugene Ng (Rice University), David G. Andersen (Carnegie Mellon
University), Michael Kaminsky and Michael A. Kozuch (Intel Labs), and Amin Vahdat (University of California, San Diego)
• データセンターで1Bpsのケーブルで100m, 10 Gpbs のケーブルで12m しか延ばせない。これを埋める
p , p
Optical networkの活用。類似の研究としてSIGCOM10のHelio/C-Throughがある。
– CloudNaaS: A Cloud Networking Platform for Enterprise Applications [full paper]
• Theophilus Benson and Aditya Akella (University of Wisconsin, Madison) and Anees Shaikh and Sambit Sahu (IBM
Research, TJ Watson)
– To Cloud Or Not To Cloud? Musings On Costs and Viability [short paper, short
presentation]
• Yao Chen and Radu Sion (Stony Brook University)
– ActiveSLA: A Profit-Oriented Admission Control Framework for Database-as-a-
Service Providers [full paper]
• Pengcheng Xiong (Georgia Institute of Technology), Yun Chi, Shenghuo Zhu, and Junichi Tatemura (NEC Laboratories
America), Calton Pu (Georgia Institute of Technology), and Hakan Hacigumus (NEC Laboratories America)
• POSTER SESSION

27. プログラム2日目 午前
• Data and How to Protect It Chair: Boon Thau Loo
– PipeCloud: Using Causality to Overcome Speed-of-Light Delays in Cloud-Based Disaster
Recovery [full paper]
• Timothy Wood (University of Massachusetts Amherst), H. Andres Lagar-Cavilla and K.K. Ramakrishnan (AT&T Research),
Prashant Shenoy (University of Massachusetts Amherst), and Jacobus Van der Merwe (AT&T Research)
– Silverline: Toward Data Confidentiality in Storage-Intensive Cloud Applications [full paper]
• Krishna Puttaswamy, Chris Kruegel, and Ben Zhao (UCSB)
– Policy expressivity in the Anzere personal cloud [full paper]
• Oriana Riva (Microsoft Research, Redmond) and Qin Yin, Dejan Juric, Ercan Ucan, and Timothy Roscoe (ETH Zurich)
– An Efficient Multi-Tier Tablet Server Storage Architecture [full paper, short presentation]
• Richard P. Spillane, Pradeep J. Shetty, Erez Zadok, Sagar Dixit, and Shrikar Archak (Stony Brook University)
• Big Data Chair: Chris Olston
g
– Small Cache, Big Effect: Provable Load Balancing for Randomly Partitioned Cluster Services [full
paper]
• Bin Fan, Hyeontaek Lim, and David G. Andersen (Carnegie Mellon University) and Michael Kaminsky (Intel Labs)
– Query Optimization for Massively Parallel Data Processing [full paper]
• Sai Wu and Feng Li (National University of Singapore), Sharad Mehrotra (University of California at Irvine), and Beng Chin Ooi
(National University of Singapore)
– Trojan Data Layouts: Right Shoes for a Running Elephant [full paper]
• Alekh Jindal, Jorge-Arnulfo Quiané-Ruiz, and Jens Dittrich (Saarland University)
– YCSB++: Benchmarking and Performance Debugging Advanced Features in Scalable Table
Stores [full paper]
• Swapnil Patil (CMU), Milo Polte (CMU), Kai Ren (CMU), Wittawat Tantisiriroj (CMU), Lin Xiao (CMU), Julio Lopez (CMU), Garth
Gibson (CMU), Adam Fuchs (National Security Agency) and Billie Rinaldi (National Security Agency)
• [SOCC10]で発表されたYCSB（Yahoo! Cloud Service Benchmark）の拡張。YCSB++はCentral
coordination とmulti-phase testを拡張。
• http://www.pdl.cmu.edu/ycsb++/

28. プログラム2日目 午後
• Managing Resources (Part I) Chair: Dongyan Xu
– Modeling and Synthesizing Task Placement Constraints in Google Compute Clusters [full paper]
• Bikash Sharma (Pennsylvania State University), Victor Chudnovsky, Joseph L. Hellerstein, and Rasekh Rifaat (Google Inc.), and
Chita R. Das (Pennsylvania State University)
– CloudScale: Elastic Resource Scaling for Multi-Tenant Cloud Systems [full paper]
• Zhiming Shen, Sethuraman Subbiah, and Xiaohui Gu (North Carolina State University) and John Wilkes (Google)
– Automatic Management of Partitioned, Replicated Search Services [short paper]
• Florian Leibert, Jake Mannix, Jimmy Lin, and Babak Hamadani (Twitter)
– DOT: A Matrix Model for Analyzing, Optimizing and Deploying Software for Big Data Analytics in
Distributed Systems [full paper]
• Yin Huai and Rubao Lee (Department of Computer Science and Engineering, The Ohio State University), Simon Zhang
(Department of Computer Science, Cornell University), and Cathy H. Xia and Xiaodong Zhang (Department of Computer Science
and Engineering, The Ohio State University)
g g, y)
• Managing Resources (Part II) Chair: Shivnath Babu
– No One (Cluster) Size Fits All: Automatic Cluster Sizing for Data-intensive Analytics [full paper]
• Herodotos Herodotou, Fei Dong, and Shivnath Babu (Duke University)
– Declarative Automated Cloud Resource Orchestration [short paper]
• Changbin Liu and Boon Thau Loo (University of Pennsylvania) and Yun Mao (AT&T Labs Research)
– CoScan: Cooperative Scan Sharing in the Cloud [full paper]
• Xiaodan Wang (Johns Hopkins University), Anish Das Sarma and Christopher Olston (Yahoo! Research), and Randal Burns
(Johns Hopkins University)
– Incoop: MapReduce for Incremental Computations [full paper]
• Pramod Bhatotia, Alexander Wieder, Rodrigo Rodrigues, Umut A. Acar, and Rafael Pasquini (MPI-SWS)

29. SOCC報告終わり

30. PLOS2011概要
• 6th Workshop on Programming Languages and Operating Systems,
October 23, 2011, Cascais, Portugal
– http://plosworkshop.org/2011/
– SOSP2011のワークショップの一つ
– 参加者 40人ぐらい
– Keynote
• The Role of Language Technology in Trustworthy Operating Systems
– Gernot Heiser, University of New South Wales and NICTA
– スライドを公開している
http://ertos.nicta.com.au/publications/papers/Heiser_11:plos.slides.pdf
– MMU-enforced protection (SeL4) v.s. Type safety (Singularity)
のコスト比較(InterDomain Comm, CrossDomain Comm,サイク
ル毎の実行コード、コード量)は面白い

31. • Static Analyses
PLOS 午前
– Finding Resource-Release Omission Faults in Linux
• Suman Saha (LIP6-Regal), Julia Lawall (DIKU, University of Copenhagen), and Gilles Muller (INRIA/LIP6-Regal)
• リソースの解放ミスはエラー処理時に起こりやすい。エラーハンドリングの終了時にリソースリリースを開
放を確認する。はLinux 2.6.34のドライバディレクトリで100個のfaultを発見。
– Configuration Coverage in the Analysis of Large-Scale System Software
• Reinhard Tartler, Daniel Lohmann, Christian Dietrich, Christoph Egger, and Julio Sincero (Friedrich-Alexander University)
• Linux 3.0 では883のKonfig fileがある。All offのKonfigでは最小カーネルにならない。
• Security
• Rounding Pointers — Type Safe Capabilities with C++ Meta Programming
• Alexander Warg and Adam Lackorzynski (Technische Universität Dresden)
• Capability = C++のsmart pointer?
– Preliminary Design of the SAFE Platform
• André DeHon, Ben Karel (University of Pennsylvania), Thomas F. Knight, Jr. (BAE Systems), Gregory Malecha (Harvard
University), Benoît Montagu (University of Pennsylvania), Robin Morisset (École Normale Supérieure Paris), Greg Morrisett
(Harvard University), Benjamin C. Pierce (University of Pennsylvania), Randy Pollack (Harvard University), Sumit Ray (BAE
Systems), Olin Shivers (Northeastern University), Jonathan M. Smith (University of Pennsylvania), and Gregory Sullivan (BAE
Systems)

32. •
PLOS 午後
Dynamic Safety and Performance
– Dynamic Deadlock Avoidance in Systems Code Using Statically Inferred Effects
• Prodromos Gerakios, Nikolaos Papaspyrou (National Technical University of Athens), Konstantinos Sagonas (National Technical
University of Athens and Uppsala University), and Panagiotis Vekris (National Technical University of Athens)
• トランザクション毎ファイルシステムの一貫性を取るRecon
データ解析言語Datalog(prolog subset)
との統合
– Using Declarative Invariants for Protecting File-System Integrity
• Jack Sun, Daniel Fryer, Ashvin Goel, and Angela Demke Brown (University of Toronto)
– Assessing the Scalability of Garbage Collectors on Many Cores
• Lokesh Gidra, Gaël Thomas, Julien Sopena, and Marc Shapiro (Regal-LIP6/INRIA)
• メニーコア上でのガベコレ評価。OpenJDKで使われているparallel copying、concurrent
Mark Sweepなどのガベコレを48コア上で評価。
• Reversible Debugging
– URDB: A Universal Reversible Debugger Based on Decomposing Debugging Histories
• Ana-Maria Visan, Kapil Arya, Gene Cooperman, and Tyler Denniston (Northeastern University)