Technological Convergence - Recently Updated Pages

Botnet masterが自爆スイッチ（kill switch)を押して１０万台のPCを道ずれに（take down)

ishida — Mon, 11 May 2009 06:59:37 CDT

Botnet master hits the kill switch, takes down 100,000 PCs
http://arstechnica.com/security/news/2009/05/zeus-botnet-hits-the-kill-switch-takes-down-100000-pcs.ars

ZeusTracker and the Nuclear Option
http://voices.washingtonpost.com/securityfix/2009/05/zeustracker_and_the_nuclear_op.html

Botnetが危険なのは大量に個人情報を盗んだり、DDoS(特定のウェブサイトをアクセス不能にする）だけではない。botに感染したPCを自爆させることも出来るのだ。最近、それが実際に発生、10万台のPCがOSを破壊された。

Zeus/Zbotに感染した１０万台の kos(kill operating system, OS(MS Windows)を自爆させる) させたBSODing の理由は現時点（２００９年５月６日）で不明。
専門家（スイス人で２１歳のRoman Hussyを含む）は以下の可能性を指摘

１．入手した金融データなどの解析のための時間稼ぎ（追跡が困難）
２．他の crime group に乗っ取られた

Zeusは極めて強力。 virus trackerで検知不能。感染マシンの収集データと共にそれに関わるサーバーの　config fileも暗号化するので、こららのマシンのgibberish-filled（訳のわからないおしゃべりに見える）fileはkey無しでは解読不能。

nVidiaによるIntelのLarrabee批判について　nVidia talks Larrabee

ishida — Sun, 19 Apr 2009 03:09:04 CDT

http://www.futuregpu.org/2009/04/nvidia-talks-larrabee.html

nVidiaによるIntelのLarrabee批判について　nVidia talks Larrabee

最近、nVidiaの著名な研究者Mr. Dally(former Chairman of Stanford University, computer science department)が１月の入社以来、沈黙を破って、Intelと、ＩｎｔｅｌのLarabeeについてコメントした記事がNew York Timesに掲載された。(＊本文は下記を参照）
http://bits.blogs.nytimes.com/2009/04/09/hello-dally-nvidia-scientist-breaks-silence-criticizes-intel/

現在、computer chip業界（Intel， AMD， ATI， nVidia）では、全ての企業がclock, power,heat(クロックスピード、消費電力、発熱)の問題と闘っている。何時の日にか、天才的なプログラマーに依存することなく、パラレルプログラミングが一般化するようなnirvana(涅槃、または、その存在そのものが現時点では不明な理想郷)に到達することを願って努力しているのだ。

Larrabeeは、このような機能（massively-parallel programming)実現のための大いなる飛躍と言える。もちろん完全ではないが。少なくとも、Little Language(CUDA, ComputeShader, OpenGL)の制約（それぞれ、API, driver, hardwareの制約が常に付きまとう）を振り払い、full C/C++のサポートをしようとする点で。

＊　April 9, 2009, 2:19 pm

Hello, Dally: Nvidia Scientist Breaks Silence, Criticizes Intel

By Ashlee Vance

My hunger strike has ended. Bill Dally is free. Back in January, Nvidia hired Mr. Dally as its new chief scientist. The move caught a lot of people’s attention, because Mr. Dally served as the chairman of Stanford University’s impressive computer science department and had done pioneering work in the semiconductor and software fields. Nvidia, however, has kept Mr. Dally under lock and key since his hiring, refusing to permit him to talk to the news media. Until now. Earlier this week, Mr. Dally and I had a chat about his decision to bolt the comfy confines of academia for Nvidia’s life-or-death struggle with Intel and Advanced Micro Devices. (Mr. Dally continues to shepherd about 12 Stanford graduate students, although he’s full time at Nvidia.) “It seemed to me that now is the time you want to be out there bringing products to the marketplace rather than writing papers,” Mr. Dally said. Mr. Dally has bought into the notion — often espoused by Nvidia’s chief executive, Jen-Hsun Huang — that we’re on the cusp of a computing revolution. Nvidia likes to say that its graphics chips will move from gaming machines and engineers’ workstations into all types of computers and servers. This assumption hinges on graphics chips’ ability to crunch through complex software faster than mainstream chips. Today’s graphics chips rely on tens and even hundreds of tiny cores that can all work together on a specific software job at the same time. They divvy up the work and then assemble the result -– a mode of computing known as parallel processing. Chips like Intel’s Core products tend to throw just a couple of very large, powerful processing cores at software. These cores are great at handling the majority of software, which runs faster when it travels through a beefier engine. The problem, however, is that Intel and A.M.D. have found it more difficult to make their big engines run faster. Instead, most of the major chip makers have called upon software writers to start chopping up their work into smaller tasks that can be spread across lots of cores if they want to see anything close to historical performance improvements. Under such a model, chips with more low-power cores will deliver the best performance, and graphics chips have a head start on multicore design. Mr. Dally envisions a world where standard chips sit alongside the graphics chips in the same computer and split up work -– something the industry has branded as heterogeneous computing. A couple of the boring, old standard chips will handle the grunt work, while lots of the graphics chips will crank away on the more demanding jobs. “Everything from people’s smartphones to laptop computers to servers will be filled with heterogeneous processor chips,” Mr. Dally said. “All of the value will be delivered by our processors.” Even today, labeling all of Nvidia’s products as graphics chips is limiting. As I wrote last year, a number of large businesses have taken Nvidia’s chips and applied them to software jobs in the oil and gas, medical and manufacturing fields. On some jobs, software gets crunched 100 to 200 times faster than with standard processors. A.M.D. has embraced the heterogeneous idea as well, although it trails Nvidia when it comes to pushing graphics chips into new markets. Intel also plans to release a radical multicore chip either this year or next. But Mr. Dally argues that Intel isn’t going radical enough with its design, code-named Larrabee, which will still rely on the company’s beloved x86 architecture. “Intel’s chip is lugging along this x86 instruction set, and there is a tax you have to pay for that,” Mr. Dally said. Intel says that staying with x86 makes life easier on software developers familiar with such an architecture. Mr. Dally rejects this by saying Intel will need to take up valuable real estate on the chip to cater to the x86 instructions. “I think their argument is mostly a marketing thing,” Mr. Dally said. Mr. Dally considered working at Intel but decided against going somewhere with what he calls a “denial architecture.” “Intel just didn’t seem like a place where I could effect very much change,” he said. “It’s so large and bureaucratic.” While Intel can afford to throw hundreds or thousands of engineers at futuristic problems, Mr. Dally will lead a team of about two dozen people. He’s trying to look 10 years out and predict areas where the company should build up its expertise. Mr. Dally is accustomed to such work, having spent years in universities trying advance the state of parallel processing far beyond the mainstream.

Big DataとMassively Parallel Computing

ishida — Thu, 05 Mar 2009 00:23:21 CST

Big DataとMassively Parallel Computing

Internet規模のBig DataのMassively Parallel 処理技術における
Googleの独占傾向の明白化とインフラオープンソース化の流れ

背景
1　2004年の米政府の”U.S. competitiveness in high performance computing"報告書
ＨＰＣ利用環境の劇的改善を要求、過去30年間の投資の成果がまったく見られないと指摘して
二つの目標実現を要求
１）ＨＰＣの具体的な実用目的への適用までの時間の劇的短縮
(特定ハードウエア用インターフェース、ドライバー開発に終始する状況からの脱却を要求)、
　　　 http://atmarkit.co.jp/news/200802/21/sgi.html
又はhttp://www.flickr.com/photos/56716141@N00/ (DSCN0024,25,27)
２）デスクトップからハイエンドまで、生産性向上実現のための共通のソフトウエア環境　
2　（インターネット規模の）”Big Data”の変換とは：Transforming Big Data http://www.hpcwire.com/features/17910714.html
Internetの普及により取り扱うデータ量は、Petabyteに達する。“Petabyte Age”では、データ量の桁違いの大きさにより、従来の常識は通用しない。情報処理に関していえば、次元における不可知論的統計学dimentionally agnostic statisticsの世界である。
Data Intensive Computingとは、データ変換システムの開発そのものである。
Supercomputingではいまや価値があるのはハードウエアではなく、それに関与する人間そのものである。
3　”Ｅｎｄ of Theory”と　Internet規模でのデータ処理におけるＲＤＢＭＳ時代の終わり　　h)　http://www.wired.com/print/science/discoveries/magazine/16-07/pb_theory
“All models are wrong, and increasingly you can succeed without them”,
an update to Google Box’s maxim
- Peter Norvig, Google’s research director, March 2008
　　(すべてのモデルは誤っており、（何らかの特定理論モデルなしでGoogleは）成功する可能性が高い。
　4　Big Data Procssing(処理・解釈)におけるGoogleアーキテクチャーの圧倒的な優位- Internet業界（含むMS, IBM）は追随に必死。
　　 Massively-parallel distributed computing,
scalability確保プロセスの自動化、
cloud computing http://network.nature.com/people/basanta/blog/2008/06/28/cloud-computing-and-science
Chubby and Paxos ((file/record lock/unlock ,failure recovery protocols in massively-parallel computing)等
http://www.hpcwire.com/topic/networks/17910529.htm
http://markets.hpcwire.com/taborcomm?Account=hpcwire&GUID=5083208&Page=MediaViewer&Ticker=PVSW

I．1998年のMicrosoft アーキテクチャー選択：なぜ、MicrosoftはYahoo買収に固執するか
II．Google の検索システムとは
III．Google 検索システムのオープンソース化
IV．Massively Parallel Computation（MPPにおけるfile close, failure recoveryメカニズム）

I. 1998年のMicrosoft アーキテクチャー選択： なぜ、MicrosoftはYahoo買収に固執するか
GoogleのMapReduceに対抗するYahooのHadoop, 検索市場の70%はGoogleが支配
http://arnoldit.com/wordpress/2008/07/08/microsoft-architecture-in-1998/
http://www.networkworld.com/community/node/30311

　Microsoft
　1）Disk IOのボトルネック（read/write時）解消に焦点 b)、
２）特定メーカーの専用ハードに固執 a)　
３）手作業でのシステムメンテナンス、スケーラビリティの実現

　Google
１）massively-parallel distributed データ処理 c)
http://tatsubori-paper.blogspot.com/
2) commodity server d)
3）スケーラビリティ自動化の壁の克服
(http://blogs.zdnet.com/BTL/?p=9027&tag=nl.e622
http://www.networkworld.com/community/node/30311
　 http://www.intelligententerprise.com/blog/archives/2008/07/two_years_to_in.html
4) Chubby (file/record lock/unlock) and Paxos (failure recovery protocols in massively-parallel computing) http://arnoldit.com/wordpress/2008/07/26/google-chubby-and-paxos

II．Google 検索システムの基幹部分
　1）Big Table,
2）Map-Reduce, 多次元データでrelational databaseに代替
3）Beehive方式、http://blogs.zdnet.com/BTL/?p=9027&tag=nl.e622
4) Chubby, Paxos　http://arnoldit.com/wordpress/2008/07/26/google-chubby-and-paxos
5）Virtualization, http://inernetnewscom/dev-news/print.php/3764026 参照
（interoperabilityとstandard architectureの実現が課題？）

III．Google 検索システムのオープンソース化 　
http://gigaom.com/2008/06/15/the-cloud-opens-up/
http://network.nature.com/blogs/user/basanta/2008/06/28/cloud-computing-and-science
1） Provisioning: Enomalism phython-based web server application, multiple hypervisor対応
２）　Ｌｉｎｋｉｎｇ　ｙｏｕｒ　ｐｒｏｇｒａｍ　ｔｏ　ｕｓｅｆｕｌ　ｓｙｓｔｅｍｓ：　Ｅｕｃａｌｙｐｔｕｓ, an elastic computing architechture
３）　Ｈａｄｏｏｐ：　Ｍａｐ－Ｒｅｄｕｃｅのopen-source版massively parallel computationデータ処理とGFS(Google File Sytstem)
４）Ｈｙｐｅｒｔａｂｌｅ：Bigtable(master, tablet server, client、分散型データベース・ストレージ、負荷分散)のＯｐｅｎ SourceＭassively parallel high performance database
http://gigazine.net/index.php?/news/comments/20080208_hypertable/
"Googleの誇る巨大データベースBigTableのオープンソースクローン「Hypertable」"
５）Puppet (Reflesh the Net), server management software that automates scaling infrastructure
http://gigaom.com/2008/06/15/the-cloud-opens-up/
６）Google 技術の個別的商品化
Ｋｉｃkｆｉｒｅ　（ＭｙＳＱＬパラレル処理専用on board cache, 専用メモリ http://blogs.zdnet.com/BTL/?p=8638）　
Aster Data (http://blogs.zdnet.com/BTL/?p=9027&tag=nl.e622
Beehive方式によるスケーラビリティ自動化、分散処理、クラウドコンピューティング　
http://network.nature.com/blogs/user/basanta/2008/06/28/cloud-computing-and-science ）
Greenplum (http://blogs.zdnet.com/Gardner/?p=2718)
MapReduceをSQL querｙと併用combine可能(Aster nCluster内で)

IV．Massively Parallel Computingの実現
１）汎用言語開発　Sawzall http://tatsubori-paper.blogspot.com , Open CL f) g)
２）Domain Speficic Parallell Programming
適用分野例
Parallel Machines社 ( http://www.parallelmachines.com)
handling in massively parallel processing
unstructured data (text, HTML, email, audio transcript),
semi-structured data (RSS, Atom, XML, JSON, OPML) ,
structured data (relational, semantic, RDF)
enterprise service bus data from a variety of sources
Microsoft 　http://www.computingatscale.com/?86
　またはhttp://www.flickr.com/photos/56716141 @N00/ (DSCN0023)
３）GPUによるDesktop HPCの実現, 価格革命と加速するmassively-parallel computing環境導入競争　http://atmarkit.co.jp/news/200802/21/sgi.html
又はhttp://www.flickr.com/photos/56716141 @N00/ (DSCN0024,25,27)
http://www.wired.com/techbiz/it/news/2008/06/gpu_power
http://www.dvhardware.net/article28316.html
http://www.pr-inside.com/print661089.htm
http://forums.vr-zone.com/showthread.php?t=317957 Nvidia: "Intel Larrabee like a GPU from 2006"
http://www.macsimumnews.com/index.php/archive/nvidia_cuda_20_released_to_production//
"Nvidia CUDA 2.0 released to production"
４）Massively parallel computing環境の導入競争（米有力研究機関、大学） e)
５） Productivity layer用Domain Specific Languageの開発に焦点
http://www.computingatscale.com/?p=75
又は http://www.flickr.com/photos/56716141@N00/ (DSCN0021,22)

注
a) clone, partition on name-brand hardware
b) farm pairs
c) in distributed setup
d) dynamic allocation
e) Nvidia : CUDA Center of Excellence at UIUC, CUDA c-compliler download, 60,000
Google/IBM: UWashington, Carnegie-Mellon, MIT, Stanford, UMaryland　での　cloud computing教育開始
Google　追従傾向の激化（MS, IBM 他）
GPUベースの massively-parallel ComputingによるDesktop HPCの実現で適用分野の急拡大
数千人のCUDA developer (oil, gas exploration, medical imaging　その他scientific research)
f) Apple, AMC, Nvidia, ARM, FreeScale, IBM, Imagination, Nokia, Motrola, Qualcomm, Samsung, TI
g) Desktop HPCの到来とOpen CLの普及でＨＰＣプロジェクトのためのスパコン使用許可、補助金申請のための国内旅行は過去のものとなるか？http://www.wired.com/techbiz/it/news/2008/06/gpu_power
h) 別紙、End of Theory 抄訳参照。Ventorの高速sequencerとHPC利用の統計的解析ツールによるショットガン遺伝子解析。
i）別紙　Transforming Big Data 抄訳参照

遺伝子geneの再定義が求められている

ishida — Wed, 04 Feb 2009 06:57:12 CST

http://www.nytimes.com/2008/11/11/science/11gene.html?_r=1&partner=rss&emc=rss&pagewanted=all

November 11, 2008
いまや、genomeの残りの部分が重大な意味を持つようになった Now: The Rest of the Genome

遺伝子geneの再定義が求められている<The gene is in an identity crisis>

DNAの大規模な研究の結果、”gene”の本質の再検討が必要となった。まさに”gene”発見100年目を前にしてgene はidentity crisisにある。例えば、DNAのある特定部分からは、異なる複数のタンパク質が生成されている可能性が明らかとなった。さらにepigenome, RNA, alien DNA、次々出現する新たな概念。

By CARL ZIMMER

より詳細な部分が重要性を増す：The Details
これまで長い間、gene とはDNAの特定の部位で、タンパク質を生成するものとの定義が行われてきた。 1968年の時点で、molecular biologist Gunther Stentと同世代の研究者達はそれ以外の詳細な部分のあるものはきわめて重要であることを理解していたが、これら例外の数々はその定義を研究する理由を正当化する程には重要であるとは思われていなかった。さらに別の問題点が1980年代、1990年代に浮かび上がってきた。RNA転写transriptである。

ゲノムとは：The Genome
2000年に初めてゲノムgenomeの大まかな分析が国際的な研究者のチームによって終了。
それ以降、研究者達はゲノムのジャングルに分け入って詳細な地図作成を開始した。
これらのプロジェクトでも大規模に行われたものにEncyclopedia of DNA Elements（Encode）がある。
Encodeの結果は以下の事実を明らかにした。すなわち、ゲノムは、少なくとも、従来想定されていたgene(遺伝子)の標準的な考えからすると極めて異様なdeeply weirdな遺伝子群（genes）で満ち溢れていることが。
いわゆる遺伝子（gene）は複数のタンパク質を生成することが出来る。代替巣プライス（接合splice）というプロセスを通じて、細胞はexonの組み合わせを変えることで異なるコピー（transcript）を生成することが出来る。Encodeチームの推定によると、タンパク質生成部分（region）は平均、5.7種類のtranscriptを生成するという。
さらに異様なこと(weirder)は、細胞は、他の遺伝子(gene)からのexonをtranscriptに投入（toss into）することがしばしば起こることである。
従って、Dr. Gingerasは“もはや、遺伝子（gene）を特定の物理的な部位にあるDNAの単一の広がりと考えることは出来ないと。
まさにgenomeの構成についての理解のパラダイムシフトであるという。

エピゲノム（後成ゲノム？）：The Epigenome
これまでのところ、ゲノムはある意味で別の形で組織化されている。すなわち、遺伝との関係で、第二のチャンネルを経由していることが判明した。
例えばtoadflax（ホソバウンラン）の白い花弁は対称性を持つが、一部は５個の斑点を持ち、それは遺伝する。しかし、DNA上の差異は無い。
実はDNAへのcap(帽子)の差異として現れる。
IDNAは各種メチル群methyl groupsに対するcapに覆われているだけではない。Histone(DNAと結合してヌクレオソームを形成するタンパク質)と呼ばれるタンパク質によっても包まれているのである。
これまでにところ、genomeと比較してepigenome 後成遺伝については、あまり良く解っていない。
これにより、ガンその他の病気の原因について、明らかになることがあるかも知れない。
Embryo胚が成長を始めるとepigenetic markは除去されるstripped away
このプロセスはきわめて微妙なものであることが判明した。
例えば1944年、オランダは大飢饉に襲われた。University of Leidenの調査により、この時期に妊娠して生まれた60人について、その兄弟と比較して、現在も、epigenetic marksがその兄弟と比較して少ないことが明らかとなった。母体がepigenetic marksの原料を供給できなかったのだ。
I少なくとも、一部について、これらの新しいepigenetic patternsは子孫に伝達されることになると思われる。
例えば、Washington State Universityで、妊娠したネズミを、ある菌類を死滅させる化学物質にさらすと、オスの精子にepigenetic markが欠落することが明らかとなった。
さらに驚くべき例として、子宮内に露出exposeされたメスは、同様の雄との交尾を避けることが同大学の研究で判明。
ただし、これらの世代を超えた変化についての重要性について、学者の意見は分かれている。

RNAが脚光を浴びる：RNA in the Spotlight
後成遺伝マークepigenetic markはその効果のみならず、まず、どのように形成されたかの点においても興味深い。例えば,DNA上のメチルグループ群methyl groupsの帽子capを被せるためには、一群のタンパク質が正確な位置spotに誘導されなければならない。研究の結果、それはその位置を見つけることが出来る何らかのRNA分子に誘導される必要があることが判明した。
これらRNAガイド群はリボゾームribosome内のRNA分子と同様に古典的な遺伝子の概念に合致しない。
これらの発見により、研究者達は人体の細胞はタンパク質を生成しないRNAをどれだけ作り出すかのか明らかにする必要に迫られている。
Encodeプロジェクトメンバーの一人のJ Matticeは次のように考えている。すなわち、これら転写の多くは、科学者達が、現在理解していない重要な多くのことimportant thingsを行っているのだと。
ルビコン川を渡った後に、後ろを振り返ると、タンパク質中心の観点はきわめて幼稚であったquite
primitiveことが解るのだと。
これらRNAを生成するgenesのある種のものは何らかの疾病の危険をもたらす可能性がある。
タンパク質を生成しないRNAの重要性にも拘わらず、Dr. BirneyはEncodeプロジェクトで発見された転写の大部分は実際にはほとんど何もしないのではないかと疑っている。
David HausslerもEncodeプロジェクトのメンバーだが、Dr Birneyと同様に“細胞はRNAを生成し、単にそれを廃棄するのだと”言う。
Dr. Hausslerはその主張の根拠を“進化evolution”に求めている。
タンパク質を生成しないDNAの４％だけが厳しい自然の選択を経験している証拠を示している。
Genomeという荷物の大部分は単に一緒に引きずっているに過ぎないと。
しかしこの荷物の中の有用と非有用の判別の基準の定義はきわめて困難である。

外来のDNA：Alien DNA
実は、ゲノムgenomeという荷物の内容の多くは死滅した遺伝子からではなく、外部から侵略してきたビールスinvading virusに由来している。
人間のゲノムは何百万年のもの時間をかけて、そこに住みついた微小なビールスの衰弱死した死骸であふれている。
これらビールスのDNAの大きな塊は時に有害な影響を及ぼすことがある。
しかし、その一部は有益な形態に進化したものもある。
この侵略するビールス群、死んでいない擬似遺伝子群pseudogenes、混ぜられたexonや後成遺伝子群はそのまま生き残ることが出来るのだろうか？
American Scientist誌の今月号（Nov.11, 2008）で、Dr. Gersteinとその生徒だったMichael Seringhausは以下のように主張する。すなわち、遺伝子geneの定義を行うにはまずRNA転写からスタートし、それをDNAに戻って追跡すべきだと。
これらの新しい概念は遺伝子をDNAの物理的な断片から離れて、より抽象的な定義に引き戻しつつある。
Gene遺伝子(概念)はその誕生から100年後の今日、再びその生まれた場所homeに戻ろうとしている。

(原文、抄訳)
November 11, 2008
いまや、genomeの残りの部分が重大な意味を持つようになった
Now: The Rest of the Genome

遺伝子geneの再定義が求められている
<The gene is in an identity crisis>

DNAの大規模な研究の結果、”gene”の本質の再検討が必要となった。まさに”gene”発見100年目を前にして。例えば、DNAのある特定部分からは、異なる複数のタンパク質が生成されている可能性が明らかとなった。

By CARL ZIMMER

Over the summer, Sonja Prohaska decided to try an experiment. She would spend a day without ever saying the word “gene.” Dr. Prohaska is a bioinformatician at the University of Leipzig in Germany. In other words, she spends most of her time gathering, organizing and analyzing information about genes. “It was like having someone tie your hand behind your back,” she said.

But Dr. Prohaska decided this awkward experiment was worth the trouble, because new large-scale studies of DNA are causing her and many of her colleagues to rethink the very nature of genes. They no longer conceive of a typical gene as a single chunk of DNA encoding a single protein. “It cannot work that way,” Dr. Prohaska said. There are simply too many exceptions to the conventional rules for genes.

It turns out, for example, that several different proteins may be produced from a single stretch of DNA. Most of the molecules produced from DNA may not even be proteins, but another chemical known as RNA. The familiar double helix of DNA no longer has a monopoly on heredity. Other molecules clinging to DNA can produce striking differences between two organisms with the same genes. And those molecules can be inherited along with DNA.

The gene, in other words, is in an identity crisis.

This crisis comes on the eve of the gene’s 100th birthday. The word was coined by the Danish geneticist Wilhelm Johanssen in 1909, to describe whatever it was that parents passed down to their offspring so that they developed the same traits. Johanssen, like other biologists of his generation, had no idea what that invisible factor was. But he thought it would be useful to have a way to describe it.

“The word ‘gene’ is completely free from any hypothesis,” Johanssen declared, calling it “a very applicable little word.”

Over the next six decades, scientists transformed that little word from an abstraction to concrete reality. They ran experiments on bread mold and bacteria, on fruit flies and corn. They discovered how to alter flowers and eyes and other traits by tinkering with molecules inside cells. They figured out that DNA was a pair of strands twisted around each other. And by the 1960s, they had a compelling definition of the gene.

A gene, they argued, was a specific stretch of DNA containing the instructions to make a protein molecule. To make a protein from a gene, a cell had to read it and build a single-stranded copy known as a transcript out of RNA. This RNA was then grabbed by a cluster of molecules called a ribosome, which used it as a template to build a protein.

A gene was also the fundamental unit of heredity. Every time a cell divided, it replicated its genes, and parents passed down some of their genes to their offspring. If you inherited red hair — or a predisposition for breast cancer — from your mother, chances were that you inherited a gene that helped produce that trait.

This definition of the gene worked spectacularly well — so well, in fact, that in 1968 the molecular biologist Gunther Stent declared that future generations of scientists would have to content themselves with “a few details to iron out.”

The Details
1968年の時点で、molecular biologist Gunther Stentと同世代の研究者達はこれらの詳細な部分のあるものはきわめて重要であることを理解していた。
Stent and his contemporaries knew very well that some of those details were pretty important. They knew that genes could be shut off and switched on when proteins clamped onto nearby bits of DNA. They also knew that a few genes encoded RNA molecules that never became proteins. Instead, they had other jobs, like helping build proteins in the ribosome.
しかしこれら例外の数々はその定義を研究する理由を正当化する程には重要であるとは思われていなかった。
But these exceptions did not seem important enough to cause scientists to question their definitions. “The way biology works is different from mathematics,” said Mark Gerstein, a bioinformatician at Yale. “If you find one counterexample in mathematics, you go back and rethink the definitions. Biology is not like that. One or two counterexamples — people are willing to deal with that.”
さらに別の問題点が1980年代、1990年代に浮かび上がってきた。RNA転写transriptである。
More complications emerged in the 1980s and 1990s, though. Scientists discovered that when a cell produces an RNA transcript, it cuts out huge chunks and saves only a few small remnants. (The parts of DNA that the cell copies are called exons; the parts cast aside are introns.) Vast stretches of noncoding DNA also lie between these protein-coding regions. The 21,000 protein-coding genes in the human genome make up just 1.2 percent of that genome.

ゲノムとは
The Genome
2000年に初めてゲノムgenomeの大まかな分析が国際的な研究者のチームによって終了。
In 2000, an international team of scientists finished the first rough draft of that genome — all of the genetic material in a human cell. They identified the location of many of the protein-coding genes, but they left the other 98.8 percent of the human genome largely unexplored.
それ以降、研究者達はゲノムのジャングルに分け入って詳細な地図作成を開始した。
Since then, scientists have begun to wade into that genomic jungle, mapping it in fine detail.
これらのプロジェクトでも大規模に行われたものにEncyclopedia of DNA Elements（Encode）がある。
One of the biggest of these projects is an effort called the Encyclopedia of DNA Elements, or Encode for short. Hundreds of scientists are carrying out a coordinated set of experiments to determine the function of every piece of DNA in the human genome. Last summer they published their results on 1 percent of the genome — some 30 million “letters” of DNA. The genetic code is written in letters, like the title of the movie “Gattaca,” with each letter standing for a molecule called a base: guanine (G), adenine (A), thymine (T), cytosine (C). The Encode team expects to have initial results on the other 99 percent by next year.
Encodeの結果は以下の事実を明らかにした。すなわち、ゲノムは、少なくとも、従来想定されていたgene(遺伝子)の標準的な考えからすると極めて異様なdeeply weirdな遺伝子群（genes）で満ち溢れていることが。
Encode’s results reveal the genome to be full of genes that are deeply weird, at least by the traditional standard of what a gene is supposed to be. “These are not oddities — these are the rule,” said Thomas R. Gingeras of Cold Spring Harbor Laboratory and one of the leaders of Encode.
いわゆる遺伝子（gene）はふくすうのタンパク質を生成することが出来る。代替巣プライス（接合splice）というプロセスを通じて、細胞はexonの組み合わせを変えることで異なるコピー（transcript）を生成することが出来る。Encodeチームの推定によると、タンパク質生成部分（region）は平均、5.7種類のtranscriptを生成するという。
A single so-called gene, for example, can make more than one protein. In a process known as alternative splicing, a cell can select different combinations of exons to make different transcripts. Scientists identified the first cases of alternative splicing almost 30 years ago, but they were not sure how common it was. Several studies now show that almost all genes are being spliced. The Encode team estimates that the average protein-coding region produces 5.7 different transcripts. Different kinds of cells appear to produce different transcripts from the same gene.
さらに異様なこと(weirder)は、細胞は、他の遺伝子(gene)からのexonをtranscriptに投入（toss into）することがしばしば起こることである。
Even weirder, cells often toss exons into transcripts from other genes. Those exons may come from distant locations, even from different chromosomes.
従って、Dr. Gingerasは“もはは遺伝子（gene）を特定の物理的な部位にあるDNA の単一の広がりと考えることは出来ないと。
So, Dr. Gingeras argues, we can no longer think of genes as being single stretches of DNA at one physical location.
まさにgenomeの構成についての理解のパラダイムシフトであるという。
“I think it’s a paradigm shift in how we think the genome is organized,” Dr. Gingeras said.
エピゲノム（後成ゲノム？）
The Epigenome
これまでのところ、ゲノムはある意味で別の形で組織化されている。すなわち、遺伝との関係で、第二のチャンネルを経由していることが判明した。
But it turns out that the genome is also organized in another way, one that brings into question how important genes are in heredity. Our DNA is studded with millions of proteins and other molecules, which determine which genes can produce transcripts and which cannot. New cells inherit those molecules along with DNA. In other words, heredity can flow through a second channel.
例えばtoadflax（ホソバウンラン）の白い花弁は対称性を持つが、一部は５個の斑点を持ち、それは遺伝する。しかし、DNA上の差異は無い。
One of the most striking examples of this second channel is a common flower called toadflax. Most toadflax plants grow white petals arranged in a mirror-like symmetry. But some have yellow five-pointed stars. These two forms of toadflax pass down their flower to their offspring. Yet the difference between their flowers does not come down to a difference in their DNA.
実はDNAへのcap(帽子)の差異として現れる。
Instead, the difference comes down to the pattern of caps that are attached to their DNA. These caps, made of carbon and hydrogen, are known as methyl groups. The star-shaped toadflax have a distinct pattern of caps on one gene involved in the development of flowers.
DNAは各種メチル群methyl groupsに対するcapに覆われているだけではない。Histone(DNAと結合してヌクレオソームを形成するタンパク質)と呼ばれるタンパク質によっても包まれているのである。
DNA is not just capped with methyl groups; it is also wrapped around spool-like proteins called histones that can wind up a stretch of DNA so that the cell cannot make transcripts from it. All of the molecules that hang onto DNA, collectively known as epigenetic marks, are essential for cells to take their final form in the body. As an embryo matures, epigenetic marks in different cells are altered, and as a result they develop into different tissues. Once the final pattern of epigenetic marks is laid down, it clings stubbornly to cells. When cells divide, their descendants carry the same set of marks. “They help cells remember what genes to keep on, and what genes can never be turned on,” said Bradley Bernstein of Harvard University.
これまでにところ、genomeと比較してepigenome 後成遺伝については、あまり良く解っていない。NIHは$190 millionのmapping epigenetic marks on DNA in different tissuesプログラムを開始した。
Scientists know much less about this “epigenome” than the genome. In September, the National Institutes of Health began a $190 million program to start mapping epigenetic marks on DNA in different tissues. “Now we can chart all these changes beyond the gene,” said Eric Richards of Cornell University.
これにより、ガンその他の病気の原因について、明らかになることがあるかも知れない。
This survey may provide clues to the origins of cancer and other diseases. It has long been known that when DNA mutates, a cell may become prone to turning cancerous. Some studies now suggest that when epigenetic marks are disturbed, cells may also be made more vulnerable to cancer, because essential genes are shut off and genes that should be shut off are turned on. What makes both kinds of changes particularly dangerous is that they are passed down from a cell to all its descendants.
Embryo胚が成長を始めるとepigenetic markは除去されるstripped away
When an embryo begins to develop, the epigenetic marks that have accumulated on both parents’ DNA are stripped away. The cells add a fresh set of epigenetic marks in the same pattern that its parents had when they were embryos.
このプロセスはきわめて微妙なものであることが判明した。
This process turns out to be very delicate. If an embryo experiences certain kinds of stress, it may fail to lay down the right epigenetic marks.
例えば1944年、オランダは大飢饉に襲われた。University of Leidenの調査により、この時期に妊娠して生まれた60人について、その兄弟と比較して、現在も、epigenetic marksがその兄弟と比較して少ないことが明らかとなった。母体がepigenetic marksの原料を供給できなかったのだ。
In 1944, for example, the Netherlands suffered a brutal famine. Scientists at the University of Leiden recently studied 60 people who were conceived during that time. In October, the researchers reported that today they still have fewer epigenetic marks than their siblings. They suggest that during the 1944 famine, pregnant mothers could not supply their children with the raw ingredients for epigenetic marks.
少なくとも、一部について、これらの新しいepigenetic patternsは子孫に伝達されることになると思われる。
In at least some cases, these new epigenetic patterns may be passed down to future generations. Scientists are debating just how often this happens. In a paper to be published next year in The Quarterly Review of Biology, Eva Jablonski and Gal Raz of Tel Aviv University in Israel assemble a list of 101 cases in which a trait linked to an epigenetic change was passed down through three generations
例えば、Washington State Universityで、妊娠したネズミを、ある菌類を死滅させる化学物質にさらすと、オスの精子にepigenetic markが欠落することが明らかとなった。
For example, Matthew Anway of Washington State University and his colleagues found that exposing pregnant rats to a chemical for killing fungus disrupted the epigenetic marks in the sperm of male embryos. The embryos developed into adult rats that suffered from defective sperm and other disorders, like cancer. The males passed down their altered epigenetic marks to their own offspring, which passed them down to yet another generation.
さらに驚くべき例として、子宮内に露出exposeされたメスは、同様の雄との交尾を避けることが同大学の研究で判明。
Last year Dr. Anway and his colleagues documented an even more surprising effect of the chemical. Female rats exposed in the womb avoided mating with exposed male rats. The scientists found this preference lasted at least three generations.
ただし、これらの世代を超えた変化についての重要性について、学者の意見は分かれている。
While these experiments are eye-opening, scientists are divided about how important these generation-spanning changes are. “There’s a lot of disagreement about whether it matters,” Dr. Richards said.

RNAが脚光を浴びる
RNA in the Spotlight
後成遺伝マークepigenetic markはその効果のみならず、まず、どのように形成されたかの点においても興味深い。例えば,DNA上のメチルグループ群methyl groupsの帽子capを被せるためには、一群のタンパク質が正確な位置spotに誘導されなければならない。研究の結果、それはその位置を見つけることが出来る何らかのRNA分子に誘導される必要があることが判明した。
Epigenetic marks are intriguing not just for their effects, but also for how they are created in the first place. To place a cap of methyl groups on DNA, for example, a cluster of proteins must be guided to the right spot. It turns out they must be led there by an RNA molecule that can find it.
これらRNAガイド群はリボゾームribosome内のRNA分子と同様に古典的な遺伝子の概念に合致しない。
These RNA guides, like the RNA molecules in ribosomes, do not fit the classical concept of the gene. Instead of giving rise to a protein, these RNA molecules immediately start to carry out their own task in the cell. Over the last decade, scientists have uncovered a number of new kinds of RNA molecules that never become proteins. (Scientists call them noncoding RNA.) In 2006, for example, Craig Mello of the University of Massachusetts and Andrew Fire of Stanford University won the Nobel Prize for establishing that small RNA molecules could silence genes by interfering with their transcription.
これらの発見により、研究者達は人体の細胞はタンパク質を生成しないRNAをどれだけ作り出すかのか明らかにする必要に迫られている。
These discoveries left scientists wondering just how much noncoding RNA our cells make. The early results of Encode suggest the answer is a lot. Although only 1.2 percent of the human genome encodes proteins, the Encode scientists estimate that a staggering 93 percent of the genome produces RNA transcripts.
Encodeプロジェクトメンバーの一人のJ Matticeは次のように考えている。すなわち、これら転写の多くは、科学者達が、現在理解していない重要な多くのことimportant thingsを行っているのだと。
John Mattick, an Encode team member at the University of Queensland in Australia, is confident that a lot of those transcripts do important things that scientists have yet to understand. “My bet is the vast majority of it — I don’t know whether that’s 80 or 90 percent,” he said.
ルビコン川を渡った後に、後ろを振り返ると、タンパク質中心の観点はきわめて幼稚であったquite
primitiveことが解るのだと。
“When you cross the Rubicon and look back, you see the protein-centric view as being quite primitive,” he said.
これらRNAを生成するgenesのある種のものは何らかの疾病の危険をもたらす可能性がある。
Certain versions of those RNA-coding genes may raise the risk of certain diseases. As part of the Encode project, scientists identified the location of variations in DNA that have been linked to common diseases like cancer. A third of those variations were far from any protein-coding gene. Understanding how noncoding RNA works may help scientists figure out how to use drugs to counteract genetic risks for diseases. “This is going to be a huge topic of research this coming decade,” said Ewan Birney, one of the leaders of the Encode project at the European Bioinformatics Institute.
タンパク質を生成しないRNAの重要性にも拘わらず、Dr. BirneyはEncodeプロジェクトで発見された転写の大部分は実際にはほとんど何もしないのではないかと疑っている。
Despite the importance of noncoding RNA, Dr. Birney suspects that most of the transcripts discovered by the Encode project do not actually do much of anything. “I think it’s a hypothesis that has to be on the table,” he said.
David HausslerもEncodeプロジェクトのメンバーだが、Dr Birneyと同様に“細胞はRNAを生成し、単にそれを廃棄するのだと”言う。
David Haussler, another Encode team member at the University of California, Santa Cruz, agrees with Dr. Birney. “The cell will make RNA and simply throw it away,” he said.
Dr. Hausslerはその主張の根拠を“進化evolution”に求めている。
Dr. Haussler bases his argument on evolution. If a segment of DNA encodes some essential molecule, mutations will tend to produce catastrophic damage. Natural selection will weed out most mutants. If a segment of DNA does not do much, however, it can mutate without causing any harm. Over millions of years, an essential piece of DNA will gather few mutations compared with less important ones.
タンパク質を制せ一子内DNAの４％だけが厳しい自然の選択を経験している証拠を示している。
Only about 4 percent of the noncoding DNA in the human genome shows signs of having experienced strong natural selection. Some of those segments may encode RNA molecules that have an important job in the cell. Some of them may contain stretches of DNA that control neighboring genes. Dr. Haussler suspects that most of the rest serve no function.
Genomeという荷物の大部分は単に一緒に引きずっているに過ぎないと。
“Most of it is baggage being dragged along,” he said.
しかしこの荷物の中の有用と非有用の判別の基準の定義はきわめて困難である。
But the line between the useless baggage and the useful DNA is hard to draw. Mutations can make it impossible for a cell to make a protein from a gene. Scientists refer to such a disabled piece of DNA as a pseudogene. Dr. Gerstein and his colleagues estimate that there are 10,000 to 20,000 pseudogenes in the human genome. Most of them are effectively dead, but a few of them may still make RNA molecules that serve an important function. Dr. Gerstein nicknames these functioning pseudogenes “the undead.”

Alien DNA
実は、ゲノムgenomeという荷物の内容の多くは死滅した遺伝子からではなく、外部から侵略してきたビールスinvading virusに由来している。
Much of the baggage in the genome comes not from dead genes, however, but from invading viruses. Viruses repeatedly infected our distant ancestors, adding their DNA to the genetic material passed down from generation to generation. Once these viruses invaded our genomes, they sometimes made new copies of themselves, and the copies were pasted in other spots in the genome. Over many generations, they mutated and lost their ability to move.
人間のゲノムは何百万年のもの時間をかけて、そこに住みついた微小なビールスの衰弱死した死骸であふれている。
“Our genome is littered with the rotting carcasses of these little viruses that have made their home in our genome for millions of years,” Dr. Haussler said.
これらビールスのDNAの大きな塊は時に有害な影響を及ぼすことがある。
As these chunks of viral DNA hop around, they can cause a lot of harm. They can disrupt the genome, causing it to stop making essential proteins. Hundreds of genetic disorders have been linked to their leaps. One of the most important jobs that noncoding RNA serves in the genome is preventing this virus DNA from spreading quickly.
しかし、その一部は有益な形態に進化したものもある。
Yet some of these invaders have evolved into useful forms. Some stretches of virus DNA have evolved to make RNA genes that our cells use. Other stretches have evolved into sites where our proteins can attach and switch on nearby genes. “They provide the raw material for innovation,” Dr. Haussler said.
この侵略するビールス群、死んでいない擬似遺伝子群pseudogenes、混ぜられたexonや後成遺伝子群はそのまま生き残ることが出来るのだろうか？
In this jungle of invading viruses, undead pseudogenes, shuffled exons and epigenetic marks, can the classical concept of the gene survive? It is an open question, one that Dr. Prohaska hopes to address at a meeting she is organizing at the Santa Fe Institute in New Mexico next March.
American Scientist誌の今月号（Nov.11, 2008）で、Dr. Gersteinとその生徒だったMichael Seringhausは以下のように主張する。すなわち、遺伝子geneの定義を行うにはまずRNA転写からスタートし、それをDNAに戻って追跡すべきだと。
In the current issue of American Scientist, Dr. Gerstein and his former graduate student Michael Seringhaus argue that in order to define a gene, scientists must start with the RNA transcript and trace it back to the DNA. Whatever exons are used to make that transcript would constitute a gene. Dr. Prohaska argues that a gene should be the smallest unit underlying inherited traits. It may include not just a collection of exons, but the epigenetic marks on them that are inherited as well.
これらの新しい概念は遺伝子をDNAの物理的な断片から離れて、より抽象的な定義に引き戻しつつある。
These new concepts are moving the gene away from a physical snippet of DNA and back to a more abstract definition. “It’s almost a recapture of what the term was originally meant to convey,” Dr. Gingeras said.
Gene遺伝子(概念)はその誕生から100年後の今日、再びその生まれた場所homeに戻ろうとしている。
A hundred years after it was born, the gene is coming home.

This article has been revised to reflect the following correction:

Correction: November 13, 2008
An article on Tuesday about new genetic research and new ideas of what a gene is misstated the number of nucleotide bases, or “letters” of DNA, that would constitute 1 percent of the human genome. One percent would be 30 million bases, or “letters,” not 3 million.

This article has been revised to reflect the following correction:

Correction: November 27, 2008
An article on Nov. 11 about a new scientific understanding of the role of genes misspelled the surname of a biologist at Washington State University who showed inherited changes in rats that seemed not to result from mutations in genes. He is Matthew Anway, not Amway.

Big DataとMassively Parallel Computing 関連記事

ishida — Fri, 16 Jan 2009 01:22:34 CST

There is no abstract available for this page revision.

parallel processing と　BI (Business Intelligence)

ishida — Fri, 16 Jan 2009 00:30:09 CST

http://www.it-director.com/business/content.php?cid=10991

parallel processing と　BI (Business Intelligence)についての技術的検討

Podcast: A technical look at parallel processing and BI
Dana Gardner By: Dana Gardner, Principal Analyst, Interarbor Solutions
Published: 8th January 2009
Copyright Interarbor Solutions © 2009

インターネット規模のデータ収集、膨大な数量のセンター出力、携帯機器群からのコンテンツによる巨大なデータの集まり、さらに企業内での分析用メタデータmetadataの恐るべき集積が従来のデータ処理モデルを破壊に瀕する限界にまで、追い詰めてしまった。
しかし、マルチコアチップセットmulti-core chipsetsを使ったパラレル処理(並行処理parallel processing)の進展はMapReduceのようなソフトウエアによる新しいアプローチを出現させた。　
ここでは、新たな又は場合によっては極限に近い各種のデータセットの要求を満たすカギは何かを考えてみたい。

BriefingsDirect's Dana Gardnerがこれらの疑問を新たなデータ構造の専門家にパネルで投げかけて、以下のポイントについて検討してもらった。具体的にはparallelism,最新データインフラ、MapReduceの統合についてである。

以下はその抜粋である。

これまで、データ量は、長年に亘りMooreの法則に従って増加を続け、さらにそれを超えた伸びを示してきた。
過去、数年間について、従来と異なるところはクロックスピードの18ヶ月ごとの倍増が停止したことである。代わりに、チップ内のcpu coreの数が増加しつつあるが、１８ヶ月ごとに倍増ということはない。
従って、データの増加はより早く、チップは基本的に静止しているが、その数は増加しているということになる。
現在、きわめて多くの人々が益々多くのデータを保存、分析しつつある
データのパラレル化（並行処理）は容易である。
必ずしも明らかでないのは、多くのデータが有効な分析を行うのに十分なスピードで分析されていないということである。
これまでに解決済みの中心的な問題点は、処理エンジンにデータを再配分し、計算処理を瞬時に行うことである。
企業は過去、１５年から２５年間、ＳＱＬに膨大な時間を費やしてきた。従って、取引データ処理システムと幅広い同時処理機能を有する多様な作業不可をSQLパラダイム基盤上のアプリケーションと一体化packageすることが極めて重要となる。
ソフトとハードの一体化については、New York Stock Exchange, Fox, MySpaceその他多くできわめて重要かつ成功裏に行われてきた。
SQLとMapReduceの統合、プログラミング環境できわめてpragmatic(実利のある)解決策を用いることで、組織内でのデータ処理能力向上が実現されることになろう。
あるデータはMacReduceでの処理が容易で、別のあるものはSQLによる処理がより容易であるということになる。
この場合、ソリューション（アプリケーション、解決策）は、ユーザがこれら全ての機能にアクセス可能にすることである。
ディベロッパーが必要とするのは、ある特定のエンジン（機能、処理、システム）がデータの配布に影響されずに、ユーザーが関心のある問題について、そのパラレル処理能力を全て使うことが出来るということである。
ある組織のデータが最終的にクラウドcloud内で最終的にどのような形に保存、処理されるかについて事前にいくつか明確にする必要がある。
まず、ビッグデータbig dataが特定の誰かに必ずしも所属せず、また、巨大であるとの前提で話を始めよう。
巨大なデータを分析するlook at作業がある場合、多数のマシンを数時間借りて、その集積されたデータpoolを圧縮すれば良い。
ディスクの密度向上スピードは低下する様子がない。
パラレル処理可能なマルチコアの計算処理コストは今後もMooreの法則に従うことになる。
このプロセスの唯一の制約はこれらシステムを使ってのプログラムを容易にし、管理可能にすることである。Cloudはシステムの管理機能の点で何らかの助けにはなるであろうし、SQL やMapReduceのプログラミング環境はパラレル処理parallelismに適合している。今後、長期に亘って膨大なデータを分析することになる。そしてデータ量は増加を続けることになる。それは成長を続けるであろう、なぜなら、それ（データ処理、分析コスト）はますます低下し、データ量は益々、増大し続けるから。

詳細はこちらのポッドキャストを参照。
Read a full transcript of the discussion. The full podcast is also available for download here.

http://briefingsdirect.blogspot.com/2009/01/technical-look-at-how-parallel.html

http://cdn4.libsyn.com/interarbor/BriefingsDirect_-_A_Technical_Discussion_on_MapReduce_and_New_Data_Architectures.mp3?nvb=20090116060507&nva=20090117061507&t=0d91530d0a5c95810400d

(原文および抄訳対照)
Podcast: A technical look at parallel processing and BI
Dana Gardner By: Dana Gardner, Principal Analyst, Interarbor Solutions
Published: 8th January 2009
Copyright Interarbor Solutions © 2009
インターネット規模のデータ収集、膨大な数量のセンター出力、携帯機器群からのコンテンツによる巨大なデータの集まり、さらに企業内での分析用メタデータmetadataの恐るべき集積が従来のデータ処理モデルを破壊に瀕する限界にまで、追い詰めてしまった。
Internet-scale data collecting, swarms of sensors outputs, and content clouds from the mobile device fabric—as well as enterprises piling up ever more kinds of analytics metadata to analyze—have stretched traditional data management models to the breaking point.
しかし、マルチコアチップセットmulti-core chipsetsを使ったパラレル処理(並行処理parallel processing)の進展はMapReduceのようなソフトウエアによる新しいアプローチを出現させた。　
Yet advances in parallel processing using multi-core chipsets have prompted new software approaches such as MapReduce that can handle these data chores at surprisingly low total cost. The technical response to oceans of data is something that has been building for some time. But the time now seems ripe to bring the technical solutions of lower-cost parallel computing advances into play with the economic imperatives of huge data crunching requirements.
ここでは、新たな又は場合によっては極限に近い各種のデータセットの要求を満たすカギは何かを考えてみたい。
And so just what are the technical underpinnings that support the new demands being placed on, and by, extreme data sets? What economies of scale can we anticipate? How will these advances spur the movement of data to Internet cloud models?
BriefingsDirect's Dana Gardnerがこれらの疑問を新たなデータ構造の専門家にパネルで投げかけて、以下のポイントについて検討してもらった。具体的にはparallelism,最新データインフラ、MapReduceの統合についてである。
BriefingsDirect's Dana Gardner put these and other questions to a panel of new data architecture experts, to plumb into how parallelism, modern data infrastructure, and MapReduce technologies come together. He spoke with Joe Hellerstein, professor of computer science at UC Berkeley; Robin Bloor, analyst at Hurwitz & Associates, and Luke Lonergan, CTO and co-founder at Greenplum.
以下はその抜粋である。
Here are some excerpts:
これまで、データ量は、長年に亘りMooreの法則に従って増加を続け、さらにそれを超えた伸びを示してきた。
Data growth has been following and exceeding Moore's Law over time. What we've been seeing is that the data sets that people are gathering and storing over time have been doubling at a rate of even faster than every 18 months. ... We're going to see all kinds of large organizations gathering data from all sorts of automated sources.
過去、数年間について、従来と異なるところはクロックスピードの18ヶ月ごとの倍増が停止したことである。代わりに、チップ内のcpu coreの数が増加しつつあるが、１８ヶ月ごとに倍増ということはない。
... What's changed in the last few years is that clock speeds on processors have stopped doubling every 18 months. ... Instead, what they are doing is putting more processing cores on every chip. You can expect the number of processors on your chip to double every 18 months, but they're not going to get any faster.
従って、データの増加はより早く、チップは基本的に静止しているが、その数は増加しているということになる。
So data is growing faster, and we have chips basically standing still, but you're getting more of them. If you want to take advantage of that data, you're going to have to program in parallel to make use of all those processors on the chips. That's the confluence that's happening.
現在、きわめて多くの人々が益々多くのデータを保存、分析しつつある
There are very many people storing and analyzing more data. We're very encouraged that most of our customers are finding new uses for data that are earning them more money. Consequently, the driver to analyze more and more data continues to grow. As our customers get more successful, this use of data is becoming really important.
データのパラレル化（並行処理）は容易である。
It's easy to parallelize the data. You break it up into little chunks and you throw it out to different machines. What can we do cleverly in computing with that kind of a framework? There are a lot of ideas for how to move forward ... where you are taking this massively parallel data-flow approach.
必ずしも明らかでないのは、多くのデータが有効な分析を行うのに十分なスピードで分析されていないということである。
One thing that's kind of invisible is that there is a lot of data out there that's not being analyzed fast enough to be analyzed effectively. That's something that I think parallelism is going to address. ... The only reason not to gather that data is when you run out of affordable processing and storage. Anybody with the budget will have as much data as they can budget for and will try to monetize that. It's going to be pervasive.
これまでに解決済みの中心的な問題点は、処理エンジンにデータを再配分し、計算処理を瞬時に行うことである。
The core problem we've solved is the ability for our engine to redistribute the data and the computation on the fly, as these queries and analysis are being performed. ... The combination of the software-switch interconnect, which Greenplum built into the Greenplum product, and the underlying use of commodity parallel computers, is brought together in this database system that makes it possible to do SQL query and languages like MapReduce with automatic parallelism.
企業は過去、１５年から２５年間、ＳＱＬに膨大な時間を費やしてきた。従って、取引データ処理システムと幅広い同時処理機能を有する多様な作業不可をSQLパラダイム基盤上のアプリケーションと一体化packageすることが極めて重要となる。
Businesses have invested a tremendous amount of their time over the last 15 to 25 years in SQL, and some of the more traditional kinds of business analysis that pay off very well are ensconced in that programming model. So, packaging a system that can do transactional, mixed workloads with large amounts of concurrency, with applications that use the SQL paradigm, is very important.
ソフトとハードの一体化については、New York Stock Exchange, Fox, MySpaceその他多くできわめて重要かつ成功裏に行われてきた。
Packaging this together as software plus hardware, making that available as a reference architecture for customers, has been very important and has been very successful in our accounts at New York Stock Exchange, Fox, MySpace, and many others.
SQLとMapReduceの統合、プログラミング環境できわめてpragmatic(実利のある)解決策を用いることで、組織内でのデータ処理能力向上が実現されることになろう。
The combination of SQL and MapReduce in a unified way in programming environments ... is a very pragmatic [step] that can help with people's ability to get their hands on data in an organization. ... You want to have the same access to all your data via either an SQL interface or a MapReduce programming interface. ... You ought to be able to access those with whatever language suits you, mix and match.
あるデータはMacReduceでの処理が容易で、別のあるものはSQLによる処理がより容易であるということになる。
Some things are easier to do in MapReduce, and some things are easier to do in SQL, even when you know both. Good programmers have a lot of tools in their tool belt. They like to be able to use whatever tool is appropriate for the task. Having both of these things interleaved is really quite helpful.
解決策としては、ユーザがこれら全ての機能にアクセス可能であるようにすることである。
[The solution] is about users being able to gain access to all that power. What really turned the corner for general data analysis using SQL is the ability for a user to not to have to worry about what kind of table structure they have. They can have lots of small tables joining to lots of big tables, and big tables joining to each other.
ディベロッパーが必要とするのは、ある特定のエンジン（機能、処理、システム）がデータの配布に影響されずに、ユーザーが関心のある問題について、そのパラレル処理能力を全て使うことが出来るということである。
What the developer needs is an engine that doesn't care how the data is distributed, per se, just being able to use all of that parallelism on the problems of interest. ... The physical model of how the database is distributed in a shared nothing architecture in a Greenplum system is not visible to the developer.
ある組織のデータが最終的にクラウドcloud内で最終的にどのような形に保存、処理されるかについて事前にいくつか明確にする必要がある。
There are a couple of questions about how an individual organization's data will end up in the cloud. Inevitably it will, but in the short-term, people like to keep their data close, particularly database data that's traditionally been in the warehouses, very carefully managed. ... It's going to be some time until we really see everybody's data warehouses up in the cloud. ... How long will it be until you really get big volumes of data in the cloud[?] The answer is that certainly new applications will be up there. We may start to see old data getting uploaded in the cloud as well.
まず、ビッグデータbig dataが特定の誰かに必ずしも所属せず、また、巨大であるとの前提で話を始めよう。
We'll start to see big data sets up there that don't necessarily belong to anyone, and they are going to be big. In that environment, you can imagine big data analytics will have to run in the cloud, because that's where the data will be. One of the fun things about the cloud that's really exciting is the elasticity of the resources. You don't buy yourself a data center full of machines, but you rent as many machines as you need for a task.
巨大なデータを分析するlook at作業がある場合、多数のマシンを数時間借りて、その集積されたデータpoolを圧縮すれば良い。
If you have a task that's going to look at a lot of data, you would rent a lot of machines for a few hours, and then you would shrink your pool. What this is going to allow people to do is that even small organizations may, for a short period of time, look at an enormous amount of data, which perhaps doesn't originate in their own data production environment, but is something that they want to utilize for their purposes.
ディスクの密度向上スピードは低下する様子がない。
Disk densities show no signs of slowing down. So, data is going to be essentially no cost. The data-gathering infrastructure is also going to be mechanized. We're going through what I call the industrial revolution of data production. We're just going to build machines to generate data, because we think we can get value out of that data, and we can store it essentially for free.
パラレル処理可能なマルチコアの計算処理コストは今後もMooreの法則に従うことになる。
The compute cost of multi-core with parallelism is going to continue Moore's Law. It's just going to continue it in a parallel programming environment. If we can get all those cores looking at all that data, it won't cost much to do that, and the cost of that will continue to shrink by half.
このプロセスの唯一の制約はこれらシステムを使ってのプログラムを容易にし、管理可能にすることである。Cloudはシステムの管理機能の点で何らかの助けにはなるであろうし、SQL やMapReduceのプログラミング環境はパラレル処理parallelismに適合している。今後、長期に亘って膨大なデータを分析することになる。そしてデータ量は増加を続けることになる。それは成長を続けるであろう、なぜなら、それ（データ処理、分析コスト）はますます低下し、データ量は益々、増大し続けるから。
The only real barrier to the process is to make those systems easy to program and manageable. Cloud helps somewhat with manageability, and programming environments like SQL and MapReduce are well-suited to parallelism. We're going to just see an enormous use of data analysis over time. It's just going to grow, because it gets cheaper and cheaper and bigger and bigger.
詳細はこちらのポッドキャストを参照。
Read a full transcript of the discussion. The full podcast is also available for download here.
http://briefingsdirect.blogspot.com/2009/01/technical-look-at-how-parallel.html
http://cdn4.libsyn.com/interarbor/BriefingsDirect_-_A_Technical_Discussion_on_MapReduce_and_New_Data_Architectures.mp3?nvb=20090116060507&nva=20090117061507&t=0d91530d0a5c95810400d

Do you agree with what Dana Gardner is saying? Perhaps you feel, or even know, different? Why not post your opinion on this issue?

Personal HPC時代の幕開け

ishida — Wed, 14 Jan 2009 00:29:27 CST

http://delicious.com/shingo2/personal-hpc

2007年6月発表のGPUベースNvidiaのTeslaの発表、更に2008年に入ってCUDAによる開発環境整備が加速するのに合わせるかのように、2008年11月以降、相次いでMicrosoft, Red Hatが、Personal HPCと銘打ったシステムを発表。

これまで、最低でも数千万から何億円の単位、さらに、network, storage, maintenance、さらに開発環境そのものが、門外漢には無縁であったHPC-High Performance Computing が一般ユーザーの手が届くところに来た。

各分野の専門家ー地質学、画像処理、医療、生物学の各分野（遺伝子解析、分子生物学など）が、手元の端末や自分で管理、運用できるシステムを使って、big data( peta byteを超えるデータ）を、massively parallel computation（例えばGPUについては１machine cycleで64以上のcomputationを行うこと）環境に基づく分析、解析、計算を行うことが出来るように現実となりつつある。デスクトップ単体の環境とcloud内のクラスター利用の違いはあるにせよ、いわばdefaultで(システムの専門知識なしで)、Google類似のデータ収集、処理環境が個人のものになりつつある。

例えば、Ｎｖｉｄｉａ　Ｔｅｓｌａの例ではCUDA環境のダウンロードは数十万、CUDA環境で開発中のdeveloperの数はすでに数万に上り、開発の分野も数十を超えて、続々その成果が報告されている。http://forums.nvidia.com/index.php?showtopic=38747

Nvidia は2007年6月のTesla　発表、続いてCUDA環境をリリース、2008年2月には一体型で960個ものcpu coreを組み込んで、通常のDeskTop PC並みの外観のＨＰＣを実現。１万ドルを切る値段で、C言語利用の１万ものthread controlが可能なparallel処理環境を実現した。まさにＨＰＣの価格破壊である。

Microsoft, Red Hatの環境も１cpuで200ドルから400ドルで、クラスター環境を利用できることで、今後１年以内に、Personal HPCによる開発例が爆発的に増えることが予想される。
Its annual subscription pricing starts at $249 a node. The Microsoft stuff lists for $475 a server.http://redhat.sys-con.com/node/698598

関連記事は下記のもの以外は、随時更新されるこちらのbookmark集を参照http://delicious.com/shingo2/personal-hpc

Red HatがLinuxベースのHPC環境を提供、MicrosoftのWindows HPC Server 2008発表に続いて。
http://redhat.sys-con.com/node/698598
Red Hat Delivers Integrated Linux-based HPC Platform
Red Hat's Delivery Is Timely Since Microsoft Is Just Pushing Out the Final Code of its Windows HPC Server 2008

http://www.moneyscience.com/Technology_News/Personal_Supercomputing_a_Reality_with_Nvidia_Tesla.html
Personal Supercomputing a Reality with Nvidia Tesla

http://www.eweek.com/c/a/IT-Infrastructure/Nvidia-Details-Personal-Supercomputer-Design-Based-on-Tesla-GPU/?kc=rss
Nvidia Details ‘Personal Supercomputer` Design Based on Tesla GPU

Personl Super-Computing時代が開幕したとするブロッグ
差出人: Miha Ahronovitz <MyInnerVo...@gmail.com>
日付: Thu, 8 Jan 2009 03:29:52 -0800 (PST)
ローカル: 2009年1月8日(木) 午後8:29
件名: Personal Super-Computing
Personal Super-Computing is a new concept, that introduces also the
subject of High Performance Computing potential to be delivered via a
cloud. The cloud success depends, at the end, on how many end users
adopt it and pay, directly and indirectly,for the services they
receive.

“Some day soon, you may use a search engine to locate every frame of
home movie in which your Grandmother appears. Soldiers may be able to
recognize buried bombs from a distance and doctors may detect early
tumors automatically.” (Texas Advanced Computing Center publication, a
Sun customer).

This will be the the Personal Super-Computing. It will transform what
we call today Super-Computing - which few people in the street
understand - into a commodity. We will get it faster if two important
players, the most relevant players, develop together the potential I
described here. A single start up can not do it alone, as it requires
both high capital investments and the reputation of a world wide
player in IT.

Botnetの攻撃で廃業したISPの話 Attack of the Bots

ishida — Thu, 11 Dec 2008 04:10:09 CST

http://www.wired.com/wired/archive/14.11/botnet.html Attack of the Bots

Attack of the Bots

The latest threat to the Net: autonomous software programs that combine forces to perpetrate mayhem, fraud, and espionage on a global scale. How one company fought the new Internet mafia – and lost.

インターネットの最新の脅威：自律的なソフトウエアプログラム群が、世界規模であるいは騒動を引き起こし、詐欺を行い、諜報活動を行う

インターネットの新たなマフィアと闘い、そして破れたある企業の話

By Scott Berinato

1.最初は単なるcongestion回線の混雑だと思われた。

2.Six Apartがこのbotnetの攻撃の対象になった。

3.これは、特に同社の２つのISPをダウンさせた。

4.Botnet業界について, botnet herder

5.Reshefのサービスは２日間ダウン、更にそのDNSサービスプロバイダも攻撃された。

6.この際、どうすべきか

7.攻撃の再開とBlue Securityの廃業

１。最初は単なるcongestion（回線の混雑）だと思われた。まず、Six Apartはそのネットワークに対するすべてのアクセスを封印した。データの一斉射撃が単一のポートに集中するだけなら、それを止めることが出来なくても、それを調査することができる。この攻撃はISPの一つであるBlue Securityをターゲットのしており、そのあおりをSix Apartは受けたことになる。

とりあえず、Blue Securityの分だけ外せばいいというわけだ。ところが、そういう訳にはいかなかった。攻撃側はすぐにSix Apart本体に関心を移し、同社のサーバーはDDoS攻撃に晒された。

２。Six Apartが、この時点で、このbotnetの攻撃の対象になってしまった。

Botsのことをよく知れば知るほど、無力感に襲われるはずだと専門家は言う。

１）ビールスのように、botsはインターネットに接続したコンピュータに自ら取り付く。

２）ビールスとの違いは、botsは外部の指令に従って動き、全体として、統制のとれた動きをする。

３）統制プログラムをC&C, Command and Control指令と統制？といい、遠くはなれてコンピュータ上に存在する。

４）botsのネットワークは90％以上がジャンクメールとして、配付されるという。

５）個々に独立したソフトウエアエージェントの統制のとれた集団はオンライン詐欺の完璧な媒介物vectorとなる。

６）グーグルは最近、これに関してbotsによるクリック詐欺（善管義務違反）で訴えられて$90百万ドルで和解した。

７）botsはさらに、キーストロークをモニターして、パスワードやその他の重要な情報を収集して、IDの窃盗、クレジットカード詐欺などに関与する。2005年には、botsがポルノに添付したspamをばらまいた。

８）更に、botを使ったDDoS（処理量を超えるトラフィックを集中させてサイトをダウンさせる）があり、Six apartの場合は、この例である。2004年、オンラインギャンブルサイトがこの被害を受けて、回復させるのに１万ドルから５万ドルを支払った。競争相手の商売をつぶすためにDDoSが行われる場合もあるという。

９）botnetは以上のどれでも、また全てを遂行できる。その上、これは拡大、性能強化を続けている。

１０）botnetは全ての予想を超えたスピードで成長を続けている。2004年には、ある大きなbotnetで例えば25千個のbotで構成されているものが一秒に5Gbitsのジャンクデータを送ることができたが、2006年の２月にStormPayオンラインサービスに向けられたDDoSは12Gbpsであり、これは同社のサーバーをハウジングしていたデータセンターをダウンさせるに十分だった。

１１）あるISPによると、１百万ノードを超えるbotnetが存在し、22-24Gbpsのデータを送信可能なものもあると言う。この規模だと、インターネットの大手の回線を不安定にする可能性があるという。

３。この攻撃は、Six Apartの２つのISPをダウンさせた。この時点までに、一見、無関係に思われる攻撃がTucows, Torontoに対して行われようとしていた。TucowsはDNS サービスのプロバイダーだが、問題のBlue SecurityのDNSプロバイダーである。

１）Blue Securityがこのような悪意の攻撃を引き起こすのは、特に驚くに当たらない。その理由は、同社が悪名高いdo-not-spam registryを運用していたことにある。その結果、spamが送られるたびに、発信者にopt-out手を引けとのメッセージであるBlue Frogを送りつけていた。

２）しかし、少なくともある、一つのspammerはこれにひるまなかった。Six Apartへの攻撃の前日のMay 1,2006、Blue Securityの客は脅しのメールを受け取り始めた。たとえば "Soon, you will be found guilty of computer crimes such as DDoS attacking of Web sites, conspiracy, and sending mass unsolicited bulk email messages for everything from Viagra to porn, as long as you continue to run Blue Frog."といったような。

３）ここで、Blue SecurityのオーナーのReshefは尻込みすることは無かった。プレスリリース彼は次のようにのべた。すなわち、同社がスパムを有効に抑止してきたこと、またこの事件が脅しと欺瞞を通じてのspammerによる無駄な企てだと。

４）攻撃側は、Blue Frog botsとやり取りするサーバー上のbotnetを解き放ってこれに応じた。数時間後にantispamサービスはダウンした。

５）Reshefは客にBlue Frogが機能しない理由を説明しようとした。Israel時間の午前1時30分、TucowsにあるBlue SecurityのDNSを同社のTypePad blogに転送するように設定変更した。

６）その結果、www.bluesecurity.comとブラウザーに入力するユーザはすべて、TypePadに誘導された。当然、このURLと使った全ての不正なトラフィックも同様である。

７）最終的に、Six Apartが直接、火事に巻き込まれることになった。

８）Reshefは攻撃の背後にいる人物を知っていると考えていた。ロシア人のスパマーで、自称PharmaMasterである。この間、chat で両者にやり取りがあったという。Reshefによると、相手は裕福で、インターネットを自分のものだと考えているという。いずれにしても、相手は憤慨しており、断固としてやり抜くつもりで、異常にtech-savvy（テクノロジーに詳しい）である。botnetsのブラックマーケットでの商売に精通していることは明らか。

４。Botnet業界について

ここで、Botnet 業界についての解説を少し。

１）麻薬業界に似ている。表通りに看板を出している訳ではない。専門家、分業化されている。producer、distributer, customer（制作、販売、顧客）でそれぞれ、何らかの形で犯罪に関与。

２）ここでいうSoftware developerソフトの開発者はbot市場におけるいわば麻薬における屋内水耕マリファナ栽培農家と同等であり、botsとC&Cプログラムを受け持つ高い技術と誇る（犯罪）仲間である。新しいプログラムを開発すると、仲間のサイトにアップするか、チャットのメッセージのattachmentとして相手に渡す。

３）この段階から、このプログラムは新世代のボットへと幅広く組み込まれることになる。毎日、何百と既存のボットに新しいプログラムが組み込まれてアップロードされているという。この業界も競争が激しく、最高の輸送力、隠蔽能力が求められるという。

４）より高い能力を持ったbotsへの需要が技術革新の推進力である。たとえばC&Cは当初は、サーバー上に存在する単一のプログラムで、指令を転送する度に、簡単に見つけることができた。今日、一個のbotnetは数多くのサーバー上に潜む多数のC&Cを内蔵している可能性がある。数分毎に新しいサーバーに司令部を移動させて、検知を排除できる。

５）この様な機能より、多様な顧客が集まってくる。組織犯罪、企業スパイ、反政府活動、当然、ハッカーやスパマーはbotnetを探し出す。更に、２０カ国以上の国家で、コンピュータ攻撃プログラムが作成中であるとSymantecの担当者はいう。核発射の暗号解析をしたければ、100万台のコンピュータをその仕事に当てればいいとbotnetの専門家が昨年、イギリスの新聞に語っている。

"If you want to break a nuclear launch code," Cybertrust's Tippett told British newspaper The Independent last year, "then set a million computers to work on it.

不吉なことに,あるボットネットは"jihad" 聖戦や "allah-akbar."アラーは偉大なりと言ったチャットチャンネルを通じて制御されているという。

" Ominously, one botnet recently detected by Arbor Networks was controlled via chat channels called "jihad" and "allah-akbar."

６）一般的にはbotnet をbot-herderボットの仲介者、技術の素人の企業家entrepreneur から借りることになる。通常、開発者とは独立している。この仲介者が直接儲けることになるが、対象はいかがわしい商売だが、botnetの販売が主である。チャットやemail経由で貸すことになる。DDoSは、例えば10,000のボットからなるボットネットで１個５セントで、非常に効果的な攻撃で、およそ500ドルと言うことになる。この件は、The trial of an LA bot-herder named Jeanson James Ancheta の事件を参照のこと。It offers a glimpse into this shadowy world. Ancheta was caught when his botnet, purportedly 100,000 PCs strong, infected machines at the Naval Air Warfare Center in China Lake, California.

５。Reshefのサービスは２日間ダウン、更にそのDNSサービスプロバイダも攻撃された。

Reshefはプレス攻勢にでたが、DNSサービスのTucows は他人の喧嘩に巻き込まれて迷惑していた。

結局、Tucowsは、この客と縁を切ることにして、Blue SecurityのDNSを停止、そのURLも停止させたdisabled.

２日後、Blue Securityはまだオンライン状態にあったが、Reshefは絶望的な状態で、Prolexicという2004年にゲームサイトでのDDoS攻撃抑止で悪名高いgained notorietyを雇った。

攻撃側は、新しい防衛線に気がつき、無制限のno-holds-barred攻撃で応じてきた。怒り狂ったようにデータがProlexicのサーバーの全てのポートを激しく攻撃した。ピークで3Gpbsに達したという。しかし、Blue Securityは生き延びていた。

６。この際、どうすべきか？

はっきり言って、うまい方法はない。

あるURLが攻撃されると、ISPは動きのないnonfunctional　アドレスにそれをしむける。その結果、犠牲者には大事なメッセージも埋もれてしまうことになる。

法律はどうか？見通しは暗い。

ボットネットに対抗するには、結局、Arbor Networksと言った民間の調査会社にまかされることになる。

ISPにC&Cを停止させるといった方法を研究している人もいるが、敵は技術的に遥かに先をいっているので無駄になると専門家はいう。

Blue Securityの数十万の客についていえば、Reshefのやり方はごく一般的なものである。しかし、人気があるのと、有効かどうかは別物である。

７。攻撃の再開とBlue Securityの廃業

Prolexic の推測は誤りであることが明らかになった。攻撃側は、あきらめず、攻撃は続いた。

あるボットネットbotnetがProlexicのDNS providerであるUltraDNSを襲った。一撃で、DDoS攻撃から守るためにProlexicに依存している企業の１５％がダウンした。

ProlexicのLaslopは”以前はインターネットをWild West開拓時代の西部辺境”と呼んだが、いまや、アルカポネのいる1920年代のシカゴのようだ”とぼやく。

いずれにしても、Blue Security のCEOにとって、対策は手遅れだった。結局、降伏してBlue Securityは廃業した。

AT FIRST, IT LOOKED LIKE typical network congestion. So the system administrators weren't too concerned when TypePad blogs and LiveJournal social networks flickered like a light bulb in a faulty socket. But 15 minutes later, at 4 pm on May 2, 2006, the sites went dark, and so did the mood at Six Apart, the company that owns them. In the blink of an eye, 10 million blogs and online communities disappeared. "It looked like the servers had freaked out," CEO Barak Berkowitz recalls. Flash floods of data thundered into one network port, stopped inexplicably, then reappeared to overwhelm another. The engineers pored over logs, desperately looking for a cause. After an agonizing hunt, they found it: a distributed denial-of-service attack, or DDoS. Six Apart's servers had been inundated with so many requests that the machines couldn't possibly process them all. It was the digital equivalent of filling a fish tank with a fire hose.

The Six Apart team sealed off nearly all access to its network. The sites came back online. Ten minutes later, they crashed again. The attacker had found an unprotected entry point and aimed a volley squarely at it.

With the data fusillade focused on a single port, the engineers could study it, even if they couldn't stop it. Soon they noticed that the malicious traffic pouring into Six Apart's network was directed to a single destination: a TypePad blog owned by a customer called Blue Security, based in Herzliya, Israel. Six Apart techs tried to reach someone there, but it was the dead of night in the Middle East. For reasons the engineers didn't yet understand, Blue Security's domain name service, or DNS – the service that translated its URL into a numerical IP address – was rerouting traffic addressed to www.bluesecurity.com straight to the TypePad blog. That is, the attacker was targeting Blue Security, but Six Apart was taking the bullet. This was good news. It meant Six Apart could quarantine Blue Security's blog until the attack ended.

But the dodge didn't help. Within minutes, the attacker turned his attention to Six Apart itself. The company's servers were slammed by a reflective DDoS attack, a technique that boosts the volume of malicious traffic by running it through a sort of Internet echo chamber. At 8 pm, the sites suddenly went dark again, almost as if someone had flipped a switch. Six Apart was dead. The bots had won.

SIX APART HAD FALLEN PREY to a botnet – a network made up of independent programs, or bots, acting in concert. Over the years, corporate IT managers have learned to firewall their networks to block unauthorized intrusion and patch their system software to keep out viruses, worms, and Trojan horses. Likewise, PC owners have installed tens of millions of personal firewalls and antivirus programs. But bots are infiltrating even protected computers, and they have quickly become a bigger threat than virulent malware like the famously destructive Melissa, I Love You, and Slammer.

"After learning about bots, you might think, 'I feel hopelessly outgunned and outmatched,'" says Peter Tippett, CTO of security consultancy Cybertrust. "You are.

"Like viruses, bots spread by installing themselves on Net-connected computers. The difference is that, while viruses act individually according to an inflexible program, bots respond to external commands and then execute coordinated attacks. The operational software, known as command and control, or C&C, resides on a remote server. Think of a botnet as a terrorist sleeper cell: Its members lurk silently within ordinary desktop computers, inert and undetected, until C&C issues orders to strike.

Bots aren't always bad. In fact, the first bots were benevolent programs used to play games or manage Internet Relay Chat channels. (Most C&Cs continue to operate via chat.) Little by little, though, the term is losing its neutrality, and the phrase malicious bot is becoming redundant.

"Bots are at the center of the undernet economy," says Jeremy Linden, until recently a researcher at Arbor Networks, which tracks and defends against botnets. "Almost every major crime problem on the Net can be traced to them.

"For instance, networks of bots distribute as much as 90 percent of all junk email, says David Dagon, a doctoral student at Georgia Tech who wrote his thesis on the topic. Earlier this year, a notorious spammer from Michigan pled guilty to using botnets to send millions of email come-ons.

Coordinated swarms of independent software agents are also the perfect vector for online fraud. They automate the processes of clicking on ads that generate per-click revenue and installing pop-up advertising engines that generate per-install revenue. Bot-driven fraud has become such a big business that Google was recently sued by class-action plaintiffs who claimed that bots, not people, had clicked on their ads. The ads were priced based on how many clicks they received; apparently competitors had hired bots to jack up the rate with an avalanche of extra clicks. Charged with negligence for failing to guard against such abuses, Google settled for $90 million.

Bots can also monitor keystrokes to collect passwords and other sensitive data for identity theft and credit card fraud. In one 2005 case, bots spread spam purporting to contain pornographic attachments. When a recipient opened the file, it installed a keystroke logger that captured, among other things, LexisNexis credentials. Using that information, the hackers compromised 300,000 accounts.

And then there are bot-driven DDoS attacks, like the one that hit Six Apart. The usual motive: extortion. In 2004, botnets attacked dozens of online gambling sites. The bookmakers were told to pay between $10,000 and $50,000 to get their sites back online. Sometimes a DDoS assault aims not at extortion but at taking out a business rival. Last summer, a 19-year-old Net entrepreneur was sentenced to 30 months in prison for directing attacks against competitors of his company, which sold athletic jerseys. And the rivalry need not involve business; it could be political. When the Web site of Connecticut senatorial candidate Joe Lieberman went down on the day of the state's Democratic primary, the culprit was widely believed to be a botnet. The perpetrator remains unknown.

A single botnet can perform any or all of these actions. Moreover, it can expand and upgrade itself. Bots scan for vulnerable computers where they can proliferate, and C&Cs update the entire network with new capabilities as they're developed. Since bots have access to the host computer's Net connection, more bots means more bandwidth, which gives the botnet more power to carry out several nefarious tasks at once.
Botnets have grown much faster than anyone imagined even a few years ago. In 2004, a large botnet contained, say, 25,000 bots capable of transmitting a combined 5 Gbits per second of junk data – enough to take a company offline. But last February, a DDoS directed at the StormPay online payment service hit 12 Gbps, enough to crash the data center that housed the company's servers. ISPs report detecting botnets larger than a million nodes and some capable of sending 22 to 24 Gbps. That's a staggering volume of data, equivalent to transmitting the contents of a large pickup truck full of books every second, or a library floor full of academic journals every minute. That kind of traffic could clog some of the Internet's largest pipes. Indeed, botnet operators have boasted that they could deep-six Google. The Net is at risk.

THE ATTACK ON SIX APART partially disabled two of the company's ISPs, which then collaborated with Six Apart to fashion a defense. Berkowitz won't reveal the details, but he calls it a force field. Whatever it was, it worked. The shield went into effect at 10 pm, about six hours after the initial assault, and the site went back online. The attacks kept coming, but the force field deflected them. It held into the next morning.

By then another, seemingly unrelated bot attack was under way. It tore into Tucows, a Toronto-based Net-services outfit, overwhelming its servers. Hundreds of thousands of Web sites were down, sites that relied on the company for email, Web hosting, spam protection, and domain name services. CEO Elliot Noss, who was accustomed to rebuffing amateur attacks, called this one "staggering."

When Six Apart's engineers learned of Tucows' fate from online news reports, they called Noss, offering to help. His assailant, they figured, was the same one who had taken out their servers the night before. After all, Tucows served as DNS provider to – who else? – Blue Security.

That Blue Security should provoke such a vicious assault wasn't too surprising. The company ran a controversial do-not-spam registry. When customers signed up, they downloaded a small software agent – as it happens, a bot. Every time unwanted messages arrived on their computers, the bot worked with Blue Security's servers to figure out where the email had come from. Then it sent an opt-out message to the spammer. The bot was called Blue Frog, after a deadly amphibian species whose skin secretes a potent toxin so predators won't eat it.

Some security experts believed that using Blue Frog – a whitehat botnet – to fight blackhat botnets lowered Blue Security to the level of base criminals. It was "the worst kind of vigilante approach," John Levine of the Coalition Against Unsolicited Commercial Email told the Associated Press in July 2005. To others, though, it was a fabulous idea. CEO Eran Reshef registered 500,000 users for the free service. He liked to boast that six of the world's top 10 spammers had stopped targeting Blue Frog subscribers. The flood of opt-out messages that Blue Security directed at spammers ate up bandwidth and processing power that otherwise would have been used to send revenue-generating spam.

But one spammer, at least, wasn't willing to back down. On May 1, the day before the attack on Six Apart, some Blue Security customers started receiving threatening emails. "Unfortunately, due to the tactics used by Blue Security, you will end up receiving this message or other nonsensical spams 20-40 times more than you would normally," one message read. Another predicted: "Soon, you will be found guilty of computer crimes such as DDoS attacking of Web sites, conspiracy, and sending mass unsolicited bulk email messages for everything from Viagra to porn, as long as you continue to run Blue Frog."

Reshef didn't flinch. He issued a press release announcing new "proof" that Blue Frog was an "effective deterrent" to spam. "This incident," he wrote, "is only a futile attempt by a degenerate spammer to fight back through intimidation and extortion."

The attacker responded by loosing a botnet on the servers that communicated with Blue Frog bots. Within hours, the antispam service was disrupted.

Reshef wanted to tell his subscribers why Blue Frog wasn't working. So at around 1:30 am in Israel (3:30 pm in San Francisco, where Six Apart is located), his company had Tucows redirect Blue Security's DNS to the company's TypePad blog. Now any user typing www.bluesecurity.com into a browser went to TypePad. And so did any malicious traffic directed at that URL, putting Six Apart directly in the line of fire.

Reshef thought he knew who was behind the attack: a Russian spammer who called himself PharmaMaster. The embattled CEO exchanged taunts and braggadocio with his anonymous assailant via chat throughout the episode. "He's been taking down lots of other sites just to get at us," Reshef told a reporter. "Blue Frog is really working. That's why he's doing this. He has a lucrative business, so he decides he owns the Internet." Whoever he was, Blue Security's enemy was angry, determined, and unusually tech-savvy. And he clearly knew how to navigate the black market for botnets.

ALTHOUGH THE BOTNET underground is intensely secretive, several researchers have managed to penetrate it. They eavesdrop on chat channels that carry bot transactions, and sometimes they get closer. Jeremy Linden and an Arbor Networks colleague, Jose Nazario, have conversed with members of the community. "The truly professional ones don't talk," Linden says, "but others make amateur mistakes."

The bot market isn't like an ad hoc street-corner bazaar of cheap handguns. It's more like the narcotics business: a highly organized subculture of people fulfilling specific functions. There are producers, distributors, and customers with varying degrees of criminal involvement.

Software developers are the bot market's equivalent of indoor hydroponic marijuana farmers – a highly technical coterie responsible for coding bots and C&Cs. When developers devise a new capability, they upload it to a clandestine Web site or pass it along as an attachment to a chat message. From there, it becomes widely incorporated into a new generation of bots."

We're seeing hundreds of new pieces of code uploaded into previously existing bots every day," says Ron Plesco, an attorney with security consultancy SRA International who also manages computer emergency readiness for the Department of Homeland Security. "It has become competitive – who can write the bot application with the best payload or best ability to hide itself?"
Demand for more-capable bots creates intense pressure for innovation. Consider C&Cs: Once, command and control was a single program sitting on a chat server, vulnerable to detection as it transmitted instructions. Today, a single botnet can contain many C&Cs hiding on many servers. It can evade detection by turning over control to a new server every few minutes. Large botnets now use tiered C&C, with different levels controlling subsets of the bot army, much like a real army.

Such capabilities attract a range of customers. Organized criminals, corporate spies, and fringe political activists, along with hackers and spammers, seek out botnets. Moreover, some 20 nations have computer attack programs under way, according to congressional testimony by Symantec security director Vincent Weafer. "If you want to break a nuclear launch code," Cybertrust's Tippett told British newspaper The Independent last year, "then set a million computers to work on it." Ominously, one botnet recently detected by Arbor Networks was controlled via chat channels called "jihad" and "allah-akbar."

Typically, customers rent a botnet from a middleman, or bot-herder, a nontechnical entrepreneur who is usually independent of the developer. Bot-herders may use the botnet directly to make money through various scams, but their primary job is marketing –renting their services via chat and email. A DDoS attack, for instance, costs roughly a nickel a bot on a 10,000-bot network – about $500 for a fairly effective assault.

The trial of an LA bot-herder named Jeanson James Ancheta offers a glimpse into this shadowy world. Ancheta was caught when his botnet, purportedly 100,000 PCs strong, infected machines at the Naval Air Warfare Center in China Lake, California. After pleading guilty to four counts of conspiracy and fraud, he was sentenced last spring to 57 months in federal prison. The indictment against him reads like a bot business primer. He perpetrated click fraud, adware fraud, and DDoS attacks on his own. He was also hired to carry out DDoS attacks against corporate targets. Ancheta tested attacks before launching them, and he gave his customers free tech support. "It's easy," he once told an accomplice during an online chat, "like slicing cheese.

"WITH HIS OWN SERVICE DOWN for almost two days and his DNS provider under attack, Reshef went on a press offensive. "It's a real war right now over who controls the Internet," he told reporters on May 3. "Blue Frog is changing the spam economy. Now one renegade is systematically wreaking havoc. People have to decide: Is this the last time we let this happen?"

Elliot Noss of Tucows didn't particularly care. He was tired of being collateral damage in someone else's fight. All he wanted was to get his company back online. Late that night – more than a day after the first attack on Blue Security and roughly 12 hours after the botnet marched on Tucows – Noss fired his customer. He suspended Blue Security's DNS, disabling its URL. This took the antispam service offline for good, and life finally returned to normal for Tucows and Six Apart.

Two days later, Blue Security was still offline, and Reshef was growing desperate. He hired a company called Prolexic, which had gained notoriety for protecting companies during a wave of DDoS attacks on gaming sites in 2004. Prolexic essentially filters Internet traffic, situating its servers between a customer and the Net at large and blocking malicious messages. If data packets shooting across the Web were fans entering a football game, Prolexic would be doing pat-downs at the gate – billions per second – and denying entry to anyone packing heat. The company took over Blue Security's URL and put it back online, scanning incoming traffic for potential threats.

The attacker quickly noticed the new defense and responded with a no-holds-barred assault. Furious sorties of data pounded Prolexic's servers on all ports. Traffic peaked at around 3 Gbps, Prolexic president Keith Laslop recalls, but Blue Security stayed up.

And with that, momentum shifted. The attacks waned as the days passed. Laslop had seen this before: Apparently the attacker was becoming bored and soon would move on.

SO WHAT'S THE BEST WAY to defend against a bot attack? There's no clear answer. Different approaches address various aspects of the problem, but a comprehensive solution has proven elusive.

ISPs usually employ a blunt instrument: If they become aware that a particular URL is under attack, they route its traffic to a nonfunctional address, forwarding all messages addressed to the victim, even legitimate communications, into oblivion. This sends the attack into a black hole, but it also takes the victim offline – and if the victim is a DNS provider, all its customers as well. Moreover, this strategy addresses only DDoS attacks. It does nothing about spam, fraud, keystroke logging, and other nefarious botnet applications.

Can law enforcement help? The outlook isn't promising. Less than 2 percent of botnet cases are reported, according to Arbor Networks research chief Danny McPherson. And it gets worse. "A lot of people think you can just report a botnet to law enforcement and they'll go investigate it," says Joe Stewart, a senior security researcher at the consultancy LURHQ. "That's not true. It has to be a high-profile case involving lots of money and collateral damage" – like the Ancheta case. Even when a case meets these criteria, the FBI and the Secret Service, the two agencies primarily responsible for electronic crime, struggle with prosecution. "The impediment to fighting botnets is international law," says David Thomas, one of the FBI's authorities on criminal computer intrusion. "It's a question of jurisdiction over 100,000 computers all over the world."

Defeating botnets is thus often left to private investigators like Arbor Networks. The company's researchers monitor chat channels to intercept bot communications and set up vulnerable machines that invite bots to proliferate. If they catch one, they reverse engineer it to see what it does and how it works. Then they alert the authorities.

Some researchers are focusing on a more aggressive approach, tracing bot traffic and then persuading the relevant ISP to disable the C&C. However, C&Cs that move every few minutes have turned such efforts into a futile game of Whac-A-Mole. The bad guys are far ahead technically, says Adam Meyers, a researcher at SRA International. Bot developers have already created exploits that take advantage of IPv6, the next-gen Internet protocol, despite the fact that it has barely been implemented yet. It's as though car thieves had lock picks for 2008-model cars today.
Judging by Blue Security's half-million subscribers, Reshef's strategy – essentially an all-out counteroffensive against spam-spewing bots – is one of the most popular. But a popular response and a viable defense are two different things. Most security experts believe that fighting bots with bots leads inevitably to escalating electronic warfare – of which the Blue Security affair is a prime example.

KEITH LASLOP AT PROLEXIC guessed wrong: The attacker didn't get bored and move on. On May 16, more than a week after the last successful assault on Blue Security, a botnet stormed one of Prolexic's DNS providers, an outfit called UltraDNS. Because all traffic destined for Prolexic's subscribers flows through the company's servers for a pat-down, its DNS is critical to keeping those companies online. So, in one blow, around 15 percent of companies that relied on Prolexic to protect them from DDoS attacks – Blue Security among them – were knocked out cold.

"We used to call the Internet a sort of Wild West," Laslop muses. "Now it's more like Chicago in the 1920s with Al Capone."

In the absence of an effective defense, researchers see two potential trajectories for the Net. It could become increasingly unsafe, making online commerce – indeed, simply logging on – less and less attractive. Alternatively, Net users might resort to ever more severe vigilante tactics. Organizations like Blue Security could build more botnets to fight botnets – a dangerous game on a global, interdependent network.

Gadi Evron, former head of Internet security for the Israeli government and now principal of consultancy Beyond Security, is determined to find a better option. Working from his home in West Jerusalem, he has become recognized as the leader of an online community of bot-busters.

"We need to start doing this as more than an afterthought," he says. Researchers, vendors, and officials must come together to build a broad defense, a combination of technical, legal, regulatory, and social fortifications capable of turning back the bot tide. Last summer, he posted an impassioned plea for such an effort on his botnet discussion board. "We have fallen too far behind for this to go on," he wrote. "All it takes is some good people to make a change."

For Blue Security and its embattled CEO, though, it was already too late. Reshef searched frantically for options, but the attacker assured him that if he put his company back online, the botnet would return and take out whatever was in its way.

So Reshef surrendered. Blue Security closed up shop.

"We cannot take the responsibility for an ever-escalating cyber war through our continued operations," read a statement issued on May 17. It was signed simply "The Blue Security Team." The missive didn't mention Reshef. Once a press hound, he denied all interview requests through a spokesperson and went into hiding. Five months later he has yet to emerge.

In the space of two weeks, an unidentified assailant had carried out a series of devastating attacks. He mowed down Six Apart, Tucows, several of their top-tier service providers, Prolexic, UltraDNS, and hundreds of thousands of Web sites, along with millions of Internet users who rely on their services.

And he put Blue Security out of business. Why? Who knows. "Someone decided to get rid of Blue Security, and he did," Evron says. "It's as simple as that."

IT系

ishida — Tue, 09 Dec 2008 07:18:26 CST

Big DataとMassively Parallel Computing

Internet規模のBig DataのMassively Parallel 処理技術における
Googleの独占傾向の明白化とインフラオープンソース化の流れ

背景
1　2004年の米政府の”U.S. competitiveness in high performance computing"報告書
ＨＰＣ利用環境の劇的改善を要求、過去30年間の投資の成果がまったく見られないと指摘して
二つの目標実現を要求
１）ＨＰＣの具体的な実用目的への適用までの時間の劇的短縮
(特定ハードウエア用インターフェース、ドライバー開発に終始する状況からの脱却を要求)、
　　　 http://atmarkit.co.jp/news/200802/21/sgi.html
又はhttp://www.flickr.com/photos/56716141@N00/tags/massivelyparallel/ (DSCN0024,25,27)
２）デスクトップからハイエンドまで、生産性向上実現のための共通のソフトウエア環境　
2　（インターネット規模の）”Big Data”の変換とは：Transforming Big Data http://www.hpcwire.com/features/17910714.html
Internetの普及により取り扱うデータ量は、Petabyteに達する。“Petabyte Age”では、データ量の桁違いの大きさにより、従来の常識は通用しない。情報処理に関していえば、次元における不可知論的統計学dimentionally agnostic statisticsの世界である。
Data Intensive Computingとは、データ変換システムの開発そのものである。
Supercomputingではいまや価値があるのはハードウエアではなく、それに関与する人間そのものである。
3　”Ｅｎｄ of Theory”と　Internet規模でのデータ処理におけるＲＤＢＭＳ時代の終わり　　h)　http://www.wired.com/print/science/discoveries/magazine/16-07/pb_theory
“All models are wrong, and increasingly you can succeed without them”,
an update to Google Box’s maxim
- Peter Norvig, Google’s research director, March 2008
　　(すべてのモデルは誤っており、（何らかの特定理論モデルなしでGoogleは）成功する可能性が高い。
　4　Big Data Procssing(処理・解釈)におけるGoogleアーキテクチャーの圧倒的な優位- Internet業界（含むMS, IBM）は追随に必死。
　　 Massively-parallel distributed computing,
scalability確保プロセスの自動化、
cloud computing http://network.nature.com/people/basanta/blog/2008/06/28/cloud-computing-and-science
Chubby and Paxos ((file/record lock/unlock ,failure recovery protocols in massively-parallel computing)等
http://www.hpcwire.com/topic/networks/17910529.htm
http://markets.hpcwire.com/taborcomm?Account=hpcwire&GUID=5083208&Page=MediaViewer&Ticker=PVSW

I．1998年のMicrosoft アーキテクチャー選択：なぜ、MicrosoftはYahoo買収に固執するか
II．Google の検索システムとは
III．Google 検索システムのオープンソース化
IV．Massively Parallel Computation（MPPにおけるfile close, failure recoveryメカニズム）

I. 1998年のMicrosoft アーキテクチャー選択： なぜ、MicrosoftはYahoo買収に固執するか
GoogleのMapReduceに対抗するYahooのHadoop, 検索市場の70%はGoogleが支配
http://arnoldit.com/wordpress/2008/07/08/microsoft-architecture-in-1998/
http://www.networkworld.com/community/node/30311

　Microsoft
　1）Disk IOのボトルネック（read/write時）解消に焦点 b)、
２）特定メーカーの専用ハードに固執 a)　
３）手作業でのシステムメンテナンス、スケーラビリティの実現

　Google
１）massively-parallel distributed データ処理 c)
http://tatsubori-paper.blogspot.com/
2) commodity server d)
3）スケーラビリティ自動化の壁の克服
(http://blogs.zdnet.com/BTL/?p=9027&tag=nl.e622
http://www.networkworld.com/community/node/30311
　 http://www.intelligententerprise.com/blog/archives/2008/07/two_years_to_in.html
4) Chubby (file/record lock/unlock) and Paxos (failure recovery protocols in massively-parallel computing) http://arnoldit.com/wordpress/2008/07/26/google-chubby-and-paxos

II．Google 検索システムの基幹部分
　1）Big Table,
2）Map-Reduce, 多次元データでrelational databaseに代替
3）Beehive方式、http://blogs.zdnet.com/BTL/?p=9027&tag=nl.e622
4) Chubby, Paxos　http://arnoldit.com/wordpress/2008/07/26/google-chubby-and-paxos
5）Virtualization, http://inernetnewscom/dev-news/print.php/3764026 参照
（interoperabilityとstandard architectureの実現が課題？）

III．Google 検索システムのオープンソース化 　
http://gigaom.com/2008/06/15/the-cloud-opens-up/
http://network.nature.com/blogs/user/basanta/2008/06/28/cloud-computing-and-science
1） Provisioning: Enomalism phython-based web server application, multiple hypervisor対応
２）　Ｌｉｎｋｉｎｇ　ｙｏｕｒ　ｐｒｏｇｒａｍ　ｔｏ　ｕｓｅｆｕｌ　ｓｙｓｔｅｍｓ：　Ｅｕｃａｌｙｐｔｕｓ, an elastic computing architechture
３）　Ｈａｄｏｏｐ：　Ｍａｐ－Ｒｅｄｕｃｅのopen-source版massively parallel computationデータ処理とGFS(Google File Sytstem)
４）Ｈｙｐｅｒｔａｂｌｅ：Bigtable(master, tablet server, client、分散型データベース・ストレージ、負荷分散)のＯｐｅｎ SourceＭassively parallel high performance database
http://gigazine.net/index.php?/news/comments/20080208_hypertable/
"Googleの誇る巨大データベースBigTableのオープンソースクローン「Hypertable」"
５）Puppet (Reflesh the Net), server management software that automates scaling infrastructure
http://gigaom.com/2008/06/15/the-cloud-opens-up/
６）Google 技術の個別的商品化
Ｋｉｃkｆｉｒｅ　（ＭｙＳＱＬパラレル処理専用on board cache, 専用メモリ http://blogs.zdnet.com/BTL/?p=8638）　
Aster Data (http://blogs.zdnet.com/BTL/?p=9027&tag=nl.e622
Beehive方式によるスケーラビリティ自動化、分散処理、クラウドコンピューティング　
http://network.nature.com/blogs/user/basanta/2008/06/28/cloud-computing-and-science ）
Greenplum (http://blogs.zdnet.com/Gardner/?p=2718)
MapReduceをSQL querｙと併用combine可能(Aster nCluster内で)

IV．Massively Parallel Computingの実現
１）汎用言語開発　Sawzall http://tatsubori-paper.blogspot.com , Open CL f) g)
２）Domain Speficic Parallell Programming
適用分野例
Parallel Machines社 ( http://www.parallelmachines.com)
handling in massively parallel processing
unstructured data (text, HTML, email, audio transcript),
semi-structured data (RSS, Atom, XML, JSON, OPML) ,
structured data (relational, semantic, RDF)
enterprise service bus data from a variety of sources
Microsoft 　http://www.computingatscale.com/?86
　または http://www.flickr.com/photos/56716141@N00/tags/massivelyparallel/
３）GPUによるDesktop HPCの実現, 価格革命と加速するmassively-parallel computing環境導入競争　http://atmarkit.co.jp/news/200802/21/sgi.html
又はhttp://www.flickr.com/photos/56716141@N00/tags/massivelyparallel/
http://www.wired.com/techbiz/it/news/2008/06/gpu_power
http://www.dvhardware.net/article28316.html
http://www.pr-inside.com/print661089.htm
http://forums.vr-zone.com/showthread.php?t=317957 Nvidia: "Intel Larrabee like a GPU from 2006"
http://www.macsimumnews.com/index.php/archive/nvidia_cuda_20_released_to_production//
"Nvidia CUDA 2.0 released to production"
４）Massively parallel computing環境の導入競争（米有力研究機関、大学） e)
５） Productivity layer用Domain Specific Languageの開発に焦点
http://www.computingatscale.com/?p=75
又は http://www.flickr.com/photos/56716141@N00/tags/massivelyparallel/
注
a) clone, partition on name-brand hardware
b) farm pairs
c) in distributed setup
d) dynamic allocation
e) Nvidia : CUDA Center of Excellence at UIUC, CUDA c-compliler download, 60,000
Google/IBM: UWashington, Carnegie-Mellon, MIT, Stanford, UMaryland　での　cloud computing教育開始
Google　追従傾向の激化（MS, IBM 他）
GPUベースの massively-parallel ComputingによるDesktop HPCの実現で適用分野の急拡大
数千人のCUDA developer (oil, gas exploration, medical imaging　その他scientific research)
f) Apple, AMC, Nvidia, ARM, FreeScale, IBM, Imagination, Nokia, Motrola, Qualcomm, Samsung, TI
g) Desktop HPCの到来とOpen CLの普及でＨＰＣプロジェクトのためのスパコン使用許可、補助金申請のための国内旅行は過去のものとなるか？http://www.wired.com/techbiz/it/news/2008/06/gpu_power
h) 別紙、End of Theory 抄訳参照。Ventorの高速sequencerとHPC利用の統計的解析ツールによるショットガン遺伝子解析。
i）別紙　Transforming Big Data 抄訳参照

Home

ishida — Mon, 19 May 2008 18:57:06 CDT

The site is dedicated to the objective of gathering information pertinent to the theme of
"The Singularity is Near: When Humans Transcend Biology"
written by by Ray Kurzweil who contends the arrival of the singularity inevitable as a result of followings;
increasingly exponential evolution of technology,
a sharp rise in computing capability,
robotics and life expectancy within the next 15 years.

(Essay for Scientific American, April 2006)
Ray Kurzweil - Reprogramming Biology
Biology is now in the early stages of an historic transition to an information science, while at the same time gaining the tools to reprogram these ancient information processes.
RNA interference (RNAi), which has emerged in the last several years, is capable of turning specific genes off.
Innovative means of adding new genes, called gene therapy, are also emerging that have overcome earlier problems with achieving precise placement of the modified genetic information.
We also have new means of activating and deactivating enzymes, the workhorses of biology.
Another important line of attack is to regrow our own cells, tissues, and even whole organs, and introduce them into our bodies without surgery.
Scientists are also applying nanotechnology to go beyond the limitations of biology. A nanoengineered device developed at Rochester University is programmed to detect the antigens specific to cancer cells.
Our ability to understand and even reprogram the brain, although in early stages, is also accelerating. We are doubling the spatial resolution of voxels (3D volumes) in brain scanning each year and the latest generation of in-vivo scanners can image individual interneuronal connections firing in real time.
Now that biology is becoming an information technology, it is subject to what I call the “law of accelerating returns.” Information technologies, including biological ones, double their price-performance and capacity in less than a year.
Sequencing DNA, for example, has come down in price by half each year from $10 per base pair in 1990 to under a penny today.
The amount of genetic data we sequenced has doubled each year. It took us 15 years to sequence HIV, but we sequenced SARS in only 31 days.

Useful links
Ray Kurzweil - Reprogramming Biology
http://hsm-science.blogspot.com/2006/05/ray-kurzweil-reprogramming-biology.html

Liveblogging Singularity Summit 2007 Saturday, Sep 8 2007
http://www.acceleratingfuture.com/michael/blog/?p=549

Liveblogging the Singularity Summit 2007 - Day Two - afternoon, September 9, 2007
http://www.davidorban.com/blog/archives/2007/09/liveblogging_th_4.html

Mutations in Moms' Genes Reveal Human Migration Through the Ages
http://www.wired.com/science/discoveries/news/2007/06/genographic_project

4d visualization of humanbody
http://zoisci.wordpress.com/2007/07/05/4d-visualization-of-human-body/

Free Podcasts of University Courses: 75 Courses and Growing
http://www.oculture.com/2007/07/free_podcasts_of_university_courses_75_courses_and_growing.html

IT業界関係者注目のニュースソース　http://www.techmeme.com/

(http://www.wired.com/techbiz/media/news/2007/05/techmeme参照）

ハーバード大学発の最新バイオアニメーションの紹介Terrific Pixar-style Harvard animation on molecular biology http://technoconvergence.wetpaint.com/page/%E3%83%8F%E3%83%BC%E3%83%90%E3%83%BC%E3%83%89%E5%A4%A7%E5%AD%A6%E7%99%BA%E3%81%AE%E6%9C%80%E6%96%B0%E3%83%90%E3%82%A4%E3%82%AA%E3%82%A2%E3%83%8B%E3%83%A1%E3%83%BC%E3%82%B7%E3%83%A7%E3%83%B3%E3%81%AE%E7%B4%B9%E4%BB%8BTerrific+Pixar-style+Harvard+animation+on+molecular+biology

”REGENERATIVE MEDICINE: 2012” video at 2007 New Yorker Conference MAY 7, 2007
http://www.newyorker.com/online/video/conference/2007/atala

Inventor Ray Kurzweil on TEDTalks

http://tedblog.typepad.com/tedblog/2006/11/inventor_ray_ku.html

KurzweillAI.net

http://www.kurzweilai.net/index.html?flash=1

Wikiversity:Main Page
http://en.wikiversity.org/wiki/Wikiversity:Main_Page

Science in the News-Harvard Med School
http://www.hms.harvard.edu/sitn/

Technology Review by MIT
http://www.technologyreview.com/BioTech/

Impact Lab
http://impactlab.com/modules.php?name=News&file=article &sid=9917

NextGenLog Next generation electronics, and related technologies, news and interviews (PST)
http://nextgenlog.blogspot.com/

Streaming Video from Algorithmic Biology 2006 Conference
http://www.calit2.net/newsroom/rss.php?id=999

NewScientist.com - Instant Expert: Brilliant Minds Forecast the Next 50 Years
http://www.newscientist.com/channel/opinion/science-forecasts

遺伝子の突然変異は自己免疫を強化させるか？Genetic Mutation May Lead To Increased Autoimmunity

ishida — Sat, 08 Mar 2008 04:05:01 CST

http://www.sciencedaily.com/releases/2008/02/080221170646.htm

Genetic Mutation May Lead To Increased Autoimmunity

免疫システムB細胞の遺伝子配列組み換えrecombining sequenceに関わるDNAの構成要素の研究：レセプター編集receptor editingプロセスのみに干渉するDNAの構成要素が明らかにされた.。

これまでにレセプター編集receptor editingが免疫システムB細胞の生成に必須であることを解明したこの研究者によると”編集プロセス”は誕生直後のリンパ球の”学習期間”であるという。
Ｂ細胞はいわば誕生間もない”若いとき”に教育を受けるのだ。

骨髄で成長する期間中、Ｂ細胞はきわめて従順だがレセプター編集receptor editingを通じて、その性格を変え、ビールスの抗体を製造するようになる。したがって、自己免疫は”編集”の欠陥であると言える。

Genetic Mutation May Lead To Increased Autoimmunity

ScienceDaily (Feb. 25, 2008) — Scientists at The Scripps Research Institute have discovered that a mutation in a known DNA recombination mechanism may result in the onset of autoimmunity and an overexpression of autoreactive antibodies—molecules that attack the host—in animal models.

The new study highlights the role of "recombining sequence," a DNA element involved in the genetic reprogramming of immune system B cells, a process called receptor editing. These new findings could point toward a possible novel therapeutic target for autoimmune diseases such as lupus.

"This is the first mutation in which we were able to cripple receptor editing but not affect other processes," said David Nemazee, a Scripps Research scientist and professor of immunology, whose laboratory conducted the study. "This produced mice more prone to autoimmunity."

Nemazee notes that the mouse recombining sequence has a human equivalent, the immunoglobulin kappa deleting element. Both are believed to play a role in eliminating or editing certain genes that encode autoreactive antibodies. In other words, both help delete parts of the immune system that could start attacking the body itself in an autoimmune reaction rather than targeting foreign invaders such as bacteria or viruses.

In the study, the mouse mutation removed the recombining sequence recombination signal, blocking gene inactivation mediated by this mechanism. As a result, in the mutant mice receptor editing and self-tolerance were impaired, in some cases leading to increased autoantibody formation.

Schooling for Lymphocytes

Nemazee, who published the first study that identified receptor editing as integral to the body's production of the immune system's B cells in 1993, described the editing process as 'schooling' for lymphocytes.

"B cells get educated early in life," Nemazee said. "As they develop in the bone marrow, they are very malleable. For example, they may start out as reactive to the host, but, through receptor editing, they can be altered so they produce viral antibodies instead. This change can lead to the elimination of the autoreactive antibodies."

When a B cell develops from stem cells in the bone marrow, it rearranges its immunoglobin gene—the gene that codes for the large receptor protein that sits on the surface of the B cell—to recognize antigen (usually part of a bacteria or virus) and the antibody specific to it (which fights off the infection).

The rearrangement involves bringing together the gene segments in various combinations. Because there are multiple copies of gene segments in the genome, this means over a million possible antibody combinations.

Receptor editing allows the immune system to correct a bad receptor rather than throwing the whole cell away. Through editing, the immune system achieves the antibody diversity at the lower cost.

B cells circulate through the blood, lymph nodes, and spleen, where they mature further. Upon recognition of their specific antigen, such as a microbe, B cells begin the process of proliferating and secreting antibodies to mark bacteria and viruses for destruction.

In autoimmune diseases, such as lupus or rheumatoid arthritis, B cells produce antibodies against the host body itself. According to Nemazee, any defect or mutation in the editing process has the potential to increase the risk of autoimmunity.

"There is evidence, although it is somewhat controversial, that in systemic lupus there is an editing defect," Nemazee said. "In our study, the mice with this mutation produced more autoreactive antibodies."

While the new study stopped short of testing whether that overexpression in turn induced autoimmune disease, some research has suggested that an increased production of these antibodies can accelerate autoimmune disease. The Scripps Research team continues to investigate.

The study will appear in the February 15, 2008, print edition of the journal Immunity. Other authors of the study, DNA Rearrangement by the Mouse Immunoglobulin Kappa Deleting Element Recombining Sequence RS Promotes Immune Tolerance and Lambda B Cell Production, are José Luis Vela, Djemel Aït-Azzouzene, Bao Hoa Duong, and Takayuki Ota of The Scripps Research Institute. The paper was part of Vela's doctoral research project.

The study was supported by the National Institutes of Health.

Adapted from materials provided by Scripps Research Institute.
Need to cite this story in your essay, paper, or report? Use one of the following formats:
APA

MLA
Scripps Research Institute (2008, February 25). Genetic Mutation May Lead To Increased Autoimmunity. ScienceDaily. Retrieved March 8, 2008, from http://www.sciencedaily.com /releases/2008/02/080221170646.htm

DNAは分子レベルでテレパシーを使って交信しているか？On Telepathy of DNA Molecules

ishida — Thu, 31 Jan 2008 05:05:52 CST

http://www.medgadget.com/archives/2008/01/on_telepathy_of_dna_molecules.html

DNAは分子レベルでテレパシーを使って交信しているか？On Telepathy of DNA Molecules

DNAの類似のセグメントは相互に相手を認識していることが明らかになった。
その際、一切のタンパク質やその他の生物学的分子の助けを借りることなく、離れた場所にある相手を相互に認識する能力があるとの Imperial College London と米国のNational Institute of Health の研究。
これまでの研究ではDNAの約１０組の塩基からなる短い鎖の間の相互の認識にタンパク質が関与していることが明らかとなっているが、何百もの塩基をもつDNAの長い鎖は相互に、全体としてタンパク質の関与なしで相互に相手を認識していることが明らかとなった。
同一の塩基の組み合わせを持ったDNA同士は異なる組み合わせを持つものと比較しておよそ２倍の確率で集まりやすいことが明らかとなった。
同一のDNAが、群れの中から求める相手を、第三者の助けなしで見つけるということは、進化における組み換えの誤りを最小にし、自然な選択、DNAの修復の問題に新たな光を投げかけることになるが、それはガン、Alzheimer's、さらに老化の原因がこのような誤りが原因と考えられているからであるという。

Researchers at Imperial College London and the National Institute of Health in the US have discovered something amazing: the ability of similar DNA segments to recognize each other:
Genes have the ability to recognise similarities in each other from a distance, without any proteins or other biological molecules aiding the process, according to new research published this week in the Journal of Physical Chemistry B. This discovery could explain how similar genes find each other and group together in order to perform key processes involved in the evolution of species.

This new study shows that genes -- which are parts of double-stranded DNA with a double-helix structure containing a pattern of chemical bases - can recognise other genes with a similar pattern of chemical bases.

This ability to seek each other out could be the key to how genes identify one another and align with each other in order to begin the process of 'homologous recombination' -- whereby two double-helix DNA molecules come together, break open, swap a section of genetic information, and then close themselves up again.

Recombination is an important process which plays a key role in evolution and natural selection, and is also central to the body's ability to repair damaged DNA. Before now, scientists have not known exactly how suitable pairs of genes find each other in order for this process to begin...
Previous studies have suggested that proteins are involved in the recognition process when it occurs between short strands of DNA which only have about 10 pairs of chemical bases. This new research shows that much longer strands of DNA with hundreds of pairs of chemical bases seem able to recognise each other as a whole without protein involvement. According to the theory, this recognition mechanism is stronger the longer the genes are.

The researchers observed the behaviour of fluorescently tagged DNA molecules in a pure solution. They found that DNA molecules with identical patterns of chemical bases were approximately twice as likely to gather together than DNA molecules with different sequences.

Professor Alexei Kornyshev from Imperial College London, one of the study's authors, explains the significance of the team's results: "Seeing these identical DNA molecules seeking each other out in a crowd, without any external help, is very exciting indeed. This could provide a driving force for similar genes to begin the complex process of recombination without the help of proteins or other biological factors. Our team's experimental results seem to support these expectations."

Understanding the precise mechanism of the primary recognition stage of genetic recombination may shed light on how to avoid or minimise recombination errors in evolution, natural selection and DNA repair. This is important because such errors are believed to cause a number of genetically determined diseases including cancers and some forms of Alzheimer's, as well as contributing to ageing. Understanding this mechanism is also essential for refining precise artificial recombination techniques for biotechnologies and gene therapies of the future.

DNAは分子レベルでテレパシーを使って交信しているか？

ishida — Wed, 30 Jan 2008 06:51:11 CST

There is no abstract available for this page revision.

前立腺がんの免疫システムへの通報はごくありふれた分子がおこなっている

ishida — Wed, 23 Jan 2008 01:55:48 CST

http://www.medicalnewstoday.com/articles/93746.php
Immune System Notified Of Prostate Cancer By Common Molecule
11 Jan 2008
免疫システムはごく普通の分子を使って、前立腺がんの腫瘍を見つけ出す。
ヒストンH4が免疫システムにどのような信号を送って腫瘍に対処するかが解明されれば前立腺の新たな治療法が見つかる可能性がある。
腫瘍細胞が何らかの表示（細胞レベル）を行って免疫システムに知らせているかがこの研究の目的である。
この研究では具体的にそのような表示をマウスの前立腺腫瘍内に発見した。
免疫細胞は事実、前立腺がんを検知することがマウス実験で確認できた。
さらに、この指標はこのような微弱な反応をはるかに容易に加速していると思われる。
マウスのガン化した前立腺は特定のレセプターを持ったT-cellを多数保有していることが判明。
ガンのマウスで、過剰なレセプターが４分の３で検出された。
すなわち、マウスの免疫システムは腫瘍の特定の表示を認識していることが明らかになった。
ところでT-cellは何を実際には見ているのだろうか。
ここでミステリーは深まった。特定のレセプターをもったT-cellのみを生産。
核からの分のみがT-cellを活性化させていることが明らかに。そこで、腫瘍の特定の部位に焦点を当てることになった。
これは驚きだったという。通常、核内のタンパク質は細胞の表面に出てくることはない。
ちなみに、T-cellは細胞の表面の分子のみを識別。
結局、通常の細胞は、ばらばらにされるとT-cellが活性化されることが明らかになった。
そこで、T-cellを活性化させる特定の核タンパク質をさがしたところそれが、H4であった。
次に、なぜ、Histonが腫瘍の表面にあるか？これが次の研究テーマである。

DISCOVER誌１２月号による2007年の科学界における驚くべき記事１００のうち２つ

ishida — Wed, 26 Dec 2007 14:30:26 CST

http://discovermagazine.com/2008/jan/year-in-science-2007

DISCOVER誌１２月号による2007年の科学界における驚くべき記事１００のうち２つ

No 9 ゲノムは個人のレベルで検証可能となる

No 52 紙でできた電池

No 9 ゲノムは個人のレベルで検証可能となる
９月４日、遺伝学者Craig Venterは自らのDNAを世界に向けて公開した。その費用は$70百万である。
Venterのこの離れ業stuntは、personalized medicine追求における画期的な年である2007年の象徴的な出来事といえる。
例えば、今年２月にFDAはMammaPrintに承認を与えたが、これは乳癌患者用の検査である。
今年は同時に、genomics(ゲノム全体の解析)に基づく多数の検査が登場した年でもある。

No 52 紙でできた電池
New YorkのRensselaerPolytechnic Instituteは黒い紙でできた電池を公開した。
この電池は軽く柔軟性に富むだけでなく、人間の血液や汗からエネルギーを引き出すことができるため、いずれ、体内に埋め込んだ小型の医療機器の電源として利用可能となると思われる。

研究チームによると、まず、シリコンの表面にカーボンナノチューブを成長させ、それを、溶解したセルロースで覆う。埋め込みナノチューブが散りばめられたセルロースは柔軟性のあるシートを作り、それを基質からはがして作られる。

ナノチューブが色を黒くするが、微量なもので、素材の９０％は店頭で売られている紙と同じだという。
驚いたことに、この電池は半分に切っても、それぞれが電池として機能し、２個の電池ができたことになるだけという。

脳の配線プログラムA Wiring Diagram of the Brain

ishida — Wed, 26 Dec 2007 14:26:00 CST

http://www.technologyreview.com/Biotech/19731/?nlid=674Monday, November 19, 2007
脳の配線プログラムA Wiring Diagram of the Brain

新たな研究分野connectomics（配線解析学？）は脳内情報処理の秘密を明らかにすることができるか？
The emerging field of connectomics could help researchers decode the brain's approach to information processing.

バイオニックハンドはここまでできる：７つの実例 Bionic hand gets thumbs-up

ishida — Wed, 26 Dec 2007 14:25:07 CST

http://news.com.com/2300-11393_3-6197171-1.html?part=rss&tag=6197171&subj=news

Bionic hand gets thumbs-up
July 18, 2007 10:00 AM PDT

Touch Bionics announced Tuesday what it calls the world's first commercially available bionic hand, a prosthetic hand that moves more naturally than traditional prostheses and can hold awkward or delicate objects. While most prostheses today can open and close, allowing people to grasp some items, they lack the detailed movements of a natural hand. The i-Limb Hand gets one step closer to that natural movement, with motors in each finger that make it possible to move individual fingers and form several different grips.
The Edinburgh, Scotland-based company also announced partial hand prostheses, called ProDigits. ProDigits can replace individual fingers or parts of a hand.
26-year-old Lindsay Block was born without a left hand and has used a prosthetic limb since she was six months old. Here, she cuts an apple with the help of her Touch Bionics i-Limb Hand.
Caption text by Jennifer Guevin

Credit: Touch Bionics

100ドルパソコンの現状：第三世界のための$150のラップトップは大論争を巻き起こすFor $150, Third-World Laptop Stirs Big Debate

ishida — Wed, 26 Dec 2007 14:22:59 CST

http://www.nytimes.com/2006/11/30/technology/30laptop.htmlhttp://www.nytimes.com/2006/11/30/technology/30laptop.html
November 30, 2006
For $150, Third-World Laptop Stirs Big Debate
100ドルパソコンの現状：第三世界のための$150のラップトップは大論争を巻き起こすFor $150, Third-World Laptop Stirs Big Debate

One Laptop Per Childプロジェクトはディスプレーのコストを＄４０ドル、消費電力を８０％カット。
現在コストは＄１５０に近づいている。
現在３百万台の仮予約があり、アフリカ、ラテンアメリカ、アジアから最低一カ国からのコミットメントが５百万台に達すれば、大量生産を始めることになっている。その時期は、予定では２００７年の中頃という。
2008年末までには、製造コストは100ドル以下に下がると予想。
インターネット接続は衛星からのダウンリンク（リビア）、既存の携帯電話網（ナイジェリア）、他の国では長距離WiFiアンテナで非都市部での接続をサポートの予定。
メッシュネットワークの機能内蔵で、0.3マイルの範囲で、マシン同士がネットワークを構成。
消費電力はわずかに２ワット、通常のラップトップは２５Wから４５Wを消費。
ただしハードディスクは付属しない。
ビデオカメラのレンズがディスプレーの右側に内蔵されている。
現在、プロトタイプの１０００台がNegroponteの手元にある。
一般消費者向けの販売予定はなし。

November 30, 2006
For $150, Third-World Laptop Stirs Big Debate
By JOHN MARKOFF

CAMBRIDGE, Mass. — When computer industry executives heard about a plan to build a $100 laptop for the developing world’s children, they generally ridiculed the idea. How could you build such a computer, they asked, when screens alone cost about $100?

Mary Lou Jepsen, the chief technologist for the project, likes to refer to the insight that transformed the machine from utopian dream to working prototype as “a really wacky idea.”

Ms. Jepsen, a former Intel chip designer, found a way to modify conventional laptop displays, cutting the screen’s manufacturing cost to $40 while reducing its power consumption by more than 80 percent. As a bonus, the display is clearly visible in sunlight.

That advance and others have allowed the nonprofit project, One Laptop Per Child, to win over many skeptics over the last two and a half years. Five countries — Argentina, Brazil, Libya, Nigeria and Thailand — have made tentative commitments to put the computers into the hands of millions of students, with production in Taiwan expected to begin by mid-2007.

The laptop does not come with a Microsoft Windows operating system or even a hard drive, and the screen is small. And the cost is now closer to $150 than $100. But the price tag, even compared with low-end $500 laptops now widely available, transforms the economic equation for developing countries.

That has not prevented the effort, conceived by Nicholas Negroponte, a prominent computer researcher, from becoming the focal point of a debate over the value of computers to both learning and economic development.

The detractors include two computer industry giants, Intel and Microsoft, pushing alternative approaches. Intel has developed a $400 laptop aimed at schools as well as an education program that focuses on teachers instead of students. And Bill Gates, Microsoft’s chairman and a leading philanthropist for the third world, has questioned whether the concept is “just taking what we do in the rich world” and assuming that that is something good for the developing world, too.

Mr. Negroponte, the founding director of the M.I.T. Media Laboratory, said he was amused by the attention his little machine was getting. It is not the first time he has been challenged for proclaiming technology’s promise.

“It’s as if people spent all of their attention focusing on Columbus’s boat and not on where he was going,” he said in an interview here. “You have to remember that what this is about is education.”

Seymour Papert, a computer scientist and educator who is an adviser to the project, has argued that if young people are given computers and allowed to explore, they will “learn how to learn.” That, Mr. Papert argues, is a more valuable skill than traditional teaching strategies that focus on memorization and testing.

The idea is also that children can take on much of the responsibility for maintaining the systems, rather than relying on or creating bureaucracies to do so.

“We believe you have to leverage the kids themselves,” Ms. Jepsen said. “They’re learning machines.” As an example, she pointed to the backlight used by the laptop. Although it is designed to last five years, if it fails it can be replaced as simply as batteries are replaced in a flashlight. It is something a child can do, she said.

That philosophy, at the heart of the project’s world view, has stirred criticism for its focus on getting equipment to students rather than issues like teacher training and curriculum.

“I think it’s wonderful that the machines can be put in the hands of children and parents, and it will have an impact on their lives if they have access to electricity,” Larry Cuban, a Stanford University education professor, said in an interview. “However, if part of their rationale is that it will revolutionize education in various countries, I don’t think it will happen, and they are naïve and innocent about the reality of formal schooling.”

The debate is certain to enter a new phase when the machines go into full-scale production by Taiwan-based Quanta Computer, the world’s second-largest laptop maker. (The manufacturer, unlike the project itself, will make a profit.) Overnight, even though it will not be available to consumers, the laptop could become the best-selling portable computer in the world.

The project now has tentative commitments for three million computers and will begin large-scale manufacturing when it reaches five million with separate commitments from at least one country each in Africa, Latin America and Asia. Based on current negotiations, Mr. Negroponte says he expects that goal to be reached by mid-2007.

It got a significant boost on Nov. 15 when the Inter-American Development Bank signed an agreement to supply both loans and grants to buy the machines.

“Several years ago, I thought it was an illusion or a utopian idea,” said Juan José Daboub, managing director of the World Bank and an independent economic-development expert. “But this is now real and encouraging.”

Mr. Negroponte said the manufacturing cost was now below $150 and that it would fall below $100 by the end of 2008.

One factor setting the project apart from earlier efforts to create inexpensive computers for education is the inclusion of a wireless network capability in each machine.

The project leaders say they will employ a variety of methods for connecting to the Internet, depending on local conditions. In some countries, like Libya, satellite downlinks will be used. In others, like Nigeria, the existing cellular data network will provide connections, and in some places specially designed long-range Wi-Fi antennas will extend the wireless Internet to rural areas.

When students take their computers home after school, each machine will stay connected wirelessly to its neighbors in a self-assembling “mesh” at ranges up to a third of a mile. In the process each computer can potentially become an Internet repeater, allowing the Internet to flow out into communities that have not previously had access to it.

“The soldiers inside this Trojan horse are children with laptops,” said Walter Bender, a computer researcher who served as director of the Media Laboratory after Mr. Negroponte and now heads software development for the laptop project.

Each machine will come with a simple mechanism for recharging itself when a standard power outlet is not available. The designers experimented with a crank, but eventually discarded that idea because it seemed too fragile. Now they have settled on several alternatives, including a foot pedal as well as a hand-pulled device that works like a salad spinner.

Ms. Jepsen’s display, which removes most of the color filters but can operate in either color or monochrome modes, has made it possible to build a computer that consumes just 2 watts of power, compared with the 25 to 45 watts consumed by a conventional laptop. The ultra-low-power operation is possible because of the lack of a hard drive (the laptop uses solid-state memory, which has no moving parts and has fallen sharply in cost) and because the Advanced Micro Devices microprocessor shuts down whenever the computer is not processing information.

The designers have also gambled in designing the laptop’s software, which is based on the freely available Linux operating system, a rival to Microsoft’s Windows. Dispensing with a traditional desktop display, the software substitutes an iconic interface intended to give students a simpler view of their programs and documents and a maplike view of other connected users nearby.

A video-camera lens sits just to the right of the display, for use in videoconferencing and taking digital still photos of reasonable quality. The computer comes with a stripped-down Web browser, a simple word processor and a number of learning programs. For e-mail, the designers intend to use Google’s Web-based Gmail service.

Only one program at a time can be viewed on the laptop because of its small 7.5-inch display.

Mr. Negroponte has been a globetrotting salesman for the project, winning Libya’s participation when he was summoned by Col. Muammar el-Qaddafi to a meeting in a desert tent on a sweltering August night. But there have also been setbacks. The Indian Education Ministry rejected a proposal to order a million computers, noting that the money could be better spent on primary and secondary education.

Mr. Negroponte said he had been re-energized by the recent arrival of the first 1,000 working prototypes. The prototypes, he said, will give him new ammunition to convince government leaders that his tiny machines can be a positive force for social development. [On a visit to Brazil on Nov. 24, Mr. Negroponte presented one of the prototypes to President Luiz Inácio Lula da Silva.]

He said a program would be created to enable those in the developed world to underwrite a laptop for a child in a designated country and to correspond with the recipient by e-mail as a sort of “glorified pen-pal program.” But however attractive the idea of a $100 or $150 laptop, he said there were no plans to make it generally available to consumers.

“They should buy Dell’s $499 laptop for now,” he said. “Ours is really designed for developing nations — dusty, dirty, no or unreliable power and so on.”

In his two decades as director of the Media Laboratory, Mr. Negroponte often faced criticism because the institution’s impressive demonstrations of technology only occasionally led to commercial applications.

“He has spent his whole career being accused of being all icing and no cake,” said Michael Hawley, a computer scientist and one of Mr. Negroponte’s former students. To that kind of scoffing, he said, the laptop’s success would be Mr. Negroponte’s best retort.

その他

ishida — Wed, 26 Dec 2007 14:22:24 CST

There is no abstract available for this page revision.

2007年の科学界における画期的発見：個人間での遺伝子の大いなる差異の発見 Science's `Breakthrough of the Year'-Human Genetic Variation 2

ishida — Wed, 26 Dec 2007 05:04:54 CST

http://www.medicalnewstoday.com/articles/92569.php 2007年の科学界における画期的発見：個人間での遺伝子の大いなる差異の発見 Science's `Breakthrough of the Year'-Human Genetic Variation 21 Dec 2007 今年は、この分野の研究者たちが、個々の人間のｹﾞﾉﾑの差異の大きさに驚く　と同時にそれらが疾病や個性に関わっていることを理解し始めた年といえる。今年になって、実施されたｹﾞﾉﾑ全体での相互関係の研究の結果、多くの病気について、新たな見方がなされるようになった。具体的には、心房細動、自己免疫、b ipolar disorder、乳癌、直腸ガン、糖尿病、心臓病、高血圧、多発性硬化症およびﾘｭｳﾏﾁといった病気についてである。２００７年においては、さらに、以下のことが判明した。すなわち、DNAの何十億ものベース（base, 底、塩基）の中で、何千から何百万ものベースについて、喪失、追加またはコピーが可能であり、それにより、数世代にわたって遺伝的特徴を変化させることが可能であると。今年度の科学上の発見で次点としてあげられるのは、細胞のプログラム更新(reprogram cells)である。日本と米国の研究者達が、６月にマウスの皮膚からiPS(induced pluripotent stem被誘導多機能幹細胞）を作成し、卵細胞、精子を含んだ体組織のあらゆる細胞生成に成功、それにより、iPSは胚性幹細胞と同等の機能を有することを証明した。細胞のプログラム更新(reprogram cells)は生物医学研究biomedical researchに全く新しい分野を開くことになると専門家はいう。今後、いくつかのハードルを超えることが必要であろうが。

Technological Convergence - Recently Updated Pages

Botnet masterが自爆スイッチ（kill switch)を押して１０万台のPCを道ずれに（take down)

nVidiaによるIntelのLarrabee批判について nVidia talks Larrabee

Hello, Dally: Nvidia Scientist Breaks Silence, Criticizes Intel

Big DataとMassively Parallel Computing

遺伝子geneの再定義が求められている

Big DataとMassively Parallel Computing 関連記事

parallel processing と BI (Business Intelligence)

Personal HPC時代の幕開け

Botnetの攻撃で廃業したISPの話 Attack of the Bots

IT系

Home

遺伝子の突然変異は自己免疫を強化させるか？Genetic Mutation May Lead To Increased Autoimmunity

DNAは分子レベルでテレパシーを使って交信しているか？On Telepathy of DNA Molecules

DNAは分子レベルでテレパシーを使って交信しているか？

前立腺がんの免疫システムへの通報はごくありふれた分子がおこなっている

DISCOVER誌 １２月号による2007年の科学界における驚くべき記事１００のうち２つ

脳の配線プログラムA Wiring Diagram of the Brain

バイオニックハンドはここまでできる：７つの実例 Bionic hand gets thumbs-up

100ドルパソコンの現状：第三世界のための$150のラップトップは大論争を巻き起こすFor $150, Third-World Laptop Stirs Big Debate

その他

2007年の科学界における画期的発見：個人間での遺伝子の大いなる差異の発見 Science's `Breakthrough of the Year'-Human Genetic Variation 2

nVidiaによるIntelのLarrabee批判について　nVidia talks Larrabee

parallel processing と　BI (Business Intelligence)

DISCOVER誌１２月号による2007年の科学界における驚くべき記事１００のうち２つ