Changes between Version 4 and Version 5 of waue/2009/1119
- Timestamp:
- Nov 19, 2009, 6:26:39 PM (15 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
waue/2009/1119
v4 v5 10 10 11 11 = 1. 在異質環境下的動態排班系統 = 12 12 13 A Dynamic MapReduce Scheduler for Heterogeneous Workloads 14 13 15 * 基於異質環境下,建立三種排班方式 14 16 * 用模擬的方式,宣稱用在hadoop後能比不用快30% 15 17 16 18 = 2. 設計一個高效能的雲端平台 = 19 17 20 An New Data Parallelism Approach with High Performace Clouds 21 18 22 * 宣稱設計更為簡化,因此效能較好 19 23 * 號稱某些case比hadoop 快兩倍 20 24 21 25 = 3. parallel closed cube 演算法 = 26 22 27 A Parallel Algorithm for Closed Cube Computation 28 23 29 * parallel closed cube 是個不容易瞭解的演算法,而作者設計了一個能用在MR平台下的parallel closed cube 演算法 24 30 * 並宣稱實驗結果有得到好處 25 31 26 32 = 4. 用雲端運算處理衛星資料 = 33 27 34 Cloud Computing for Satellite Data Processing on High End Compute Clusters 35 28 36 * 用高檔設備透過Hadoop處理衛星資料 29 37 * 此篇數據比較了 有用MapReduce 以及沒用的差別 (作者說程式沒有差很多) 30 38 31 39 = 5. 一個整合計算與資料管理的系統 = 40 32 41 Clustera: An Integrated Computation And Data Management System 42 33 43 * 介紹一個資料管理系統,提供兩個特點 34 44 * 特點一為有延展性並且有能力於掌控大範圍Job 資料,並用最小的sql查詢語法減少I/O … … 37 47 38 48 = 6.用sector做高效能資料探勘 = 49 39 50 Data Mining Using High Performance Data Clouds 40 51 41 52 = 7.探勘日誌來偵測大範圍的系統問題 = 53 42 54 Detecting Large-Scale System Problems by Mining Console Logs 55 43 56 * 透過探勘log檔,來偵查出有可能出現的系統runtime problem 44 57 * 實驗於 Hadoop日誌與DarkStar線上遊戲 … … 47 60 = 8. Disco 的實驗論文 = 48 61 62 DisCo: Distributed Co-clustering with Map-Reduce 63 49 64 = 9. Sphere 的論文 = 50 51 = 10. 用Hadoop來算使用者習慣= 65 66 Exploring Data Parallelism and Locality in Wide Area Networks 67 68 = 10. 用Hadoop來算使用者習慣 = 69 52 70 Extraction of User Profile Based on the Hadoop 71 53 72 * 架構與icas很相向,果然是阿路仔 54 73 * 此篇用hadoop 來找使用者的習慣,其實只有在做map reduce 字數統計而已 55 74 * 也有畫出單一台與多台hadoop的效能比較,由於他們只有80MB的資料,因此一台最快,三台最慢 75 * Wireless Communications, Networking and Mobile Computing, 2009. WiCom '09. 5th International Conference on 56 76 57