= 2009-09-17 = * [http://rsms2.twaren.net/twaren2009/Default.aspx TWAREN 技術小組會議暨應用研討會] * 15:00-15:20 自由軟體 DRBL / Clonezilla 介紹 (國網中心.蕭志榥 組長) * 16:00-16:30 主任中群業務訪談 == Cloud Computing == * [http://www.snia.org/events/storage-developer2009/images/SDCConferenceGuide09web.pdf SNIA Releases new Cloud Data Management Interface (CDMI) for public review] (PDF) * [ Cloud Data Management Interface (CDMI) v0.80 DRAFT] (PDF) * [http://ogf.org/Resources/documents/CloudStorageForCloudComputing.pdf New White paper available on Cloud Storage for Cloud Computing SNIA/OCCI Collaboration] * [http://lib.stanford.edu/files/see_pasig_dic.pdf Data Intensive Computing] * 這個投影片留在桌面上許久了,先前就已經看過,今天稍微整理一下裡面我覺得比較重要的幾張投影片。裡面還有提到 Pig 跟 Hadoop 開發的比較,有興趣的可以下載看一下。 * 資料爆炸是這一波運算改革的主要動力,半導體產業信奉的 Moore's Law 是每十八個月電晶體數量成長兩倍。但這裡秀出很可怕的是基因資料庫資料成長的速度是每十八個月八倍。[[BR]][[Image(wiki:jazz/09-09-17:data_growth.png,width=600)]] * 一年產生出來的資料合計是 27 萬 Peta Bytes[[BR]]其中排行前三大是人類基因資料庫(7000PB),個人電子相簿(1000PB+),年度電子信箱流量(300PB+)[[BR]][[Image(wiki:jazz/09-09-17:data_types_size.png,width=600)]] * Data Intensive Computing 目前的速度瓶頸目前仍舊受限於 Disk I/O,因此除了分散式儲存外,從計算機的結構面來看,改用 SSD 甚至用 RAM Disk 是可以達成某種程度的加速。[[BR]][[Image(wiki:jazz/09-09-17:latency_comparison.png,width=600)]] * [http://udn.com/NEWS/FINANCE/FIN3/5137898.shtml 廣達 投入雲端運算] - the need of localization 本土化的需求 (讚!) {{{ 廣達研究院最近啟動「Qmulus」計畫,名字取自積層雲(Cumulous Cloud)的英文諧音, 希望透過積層雲又稱「祥瑞的雲」的象徵意涵,宣誓廣達投入雲端運算領域的決心。 廣達研究院正在思考,針對特定區域、特定應用領域,是否也可以發展出一些雲端機會,例如針對亞洲市場開發出區域雲端。 }}} == Purpose-Driven Cloud == * [http://gevaperry.typepad.com/main/2009/08/the-purposedriven-cloud.html The Purpose-Driven Cloud: Introduction] * [http://gevaperry.typepad.com/main/2009/08/the-purposedriven-cloud-part-1-usability.html Usability-Driven Clouds] * [http://gevaperry.typepad.com/main/2009/09/the-purposedriven-cloud-part-2-technology-frameworks.html Framework-Based Clouds] * Domain-Specific Clouds * Data-Driven Clouds * Industry-Focused Clouds * Geographically-Centric Clouds == ScaleDB == * [http://scaledb.com/blog/ScaleDB_Hadoop_Mapreduce_Oh_My.html Is ScaleDB Using MapReduce? Competing with Hadoop?] * [http://www.scaledb.com/ ScaleDB] frees your database from specific servers and enables you to run multiple database nodes in a cluster, with each node having full read/write access across the entire database. * 看起來 ScaleDB 是可以讓 MySQL 跑在多台上。 == HadoopDB == * [http://db.cs.yale.edu/hadoopdb/hadoopdb.html HadoopDB 官方網站] * [http://dbmsmusings.blogspot.com/2009/07/announcing-release-of-hadoopdb-longer.html HadoopDB 官方部落格] * [http://www.h-online.com/open/news/print/113822 HadoopDB reconciles SQL with Map/Reduce] * [http://www.computerworld.com/s/article/print/9135726/Yale_researchers_create_database_Hadoop_hybrid Yale researchers create database-Hadoop hybrid] * [http://radar.oreilly.com/print/37596.html HadoopDB: An Open Source Parallel Database] * [http://www.ddj.com/article/printableArticle.jhtml?articleID=219700568&dept_url=/hpc-high-performance-computing/ Introducing HadoopDB] - A hybrid of DBMS and MapReduce technologies that targets analytical workloads * [http://www.ebizq.net/blogs/enterprise/2009/09/10_ways_to_complement_the_ente.php 10 Ways To Complement the Enterprise RDBMS Using Hadoop] == Hadoop IDE : Hadoop Studio == * 除了 Eclipse 外,開發 Hadoop Map/Reduce 程式也可以用 Netbean * [註] Netbean 也有 [http://packages.debian.org/netbean Debian] / [http://packages.ubuntu.com/netbean Ubuntu 套件] * [http://www.hadoopstudio.org/ Hadoop Studio] is a Hadoop map-reduce development environment (IDE) based on NetBeans. == Hadoop == * [http://www.ebizq.net/blogs/enterprise/2009/09/10_ways_to_complement_the_ente.php 10 Ways To Complement the Enterprise RDBMS Using Hadoop] * [[Image(http://www.ebizq.net/blogs/enterprise/images/mapreduce_hadoop.png)]] == Smart Grid == * [http://www.bnext.com.tw/?Mod=PrintArticle&Func=FocusBlock&Id=2354 【新商機】資策會MIC:智慧電網商機 2012年上看600億元] {{{ 智慧電網的重要功能―能源管理系統(EMS,Energy Management System),將能幫助每個家庭了解自己的用電狀況, 並分析、建議各項省電辦法;甚至,系統還能自動調整空調的使用時間與溫度,幫助每個家庭控制用電量。 }}}