{"id":2143,"date":"2014-04-02T09:29:55","date_gmt":"2014-04-02T01:29:55","guid":{"rendered":"http:\/\/www.banbanli.net\/blog\/?p=2143"},"modified":"2014-04-02T09:29:55","modified_gmt":"2014-04-02T01:29:55","slug":"%e8%bd%89%e8%b2%bcgoogle-%e5%a4%a7%e7%a5%9e%e5%9c%a8-big-data-%e6%87%89%e7%94%a8%e4%b8%8a%e7%9a%84%e6%8a%80%e8%a1%93%e8%88%87%e8%ab%96%e6%96%87%e4%bb%8b%e7%b4%b9%ef%bc%8c%e5%a5%bd%e7%a5%9e%ef%bc%81","status":"publish","type":"post","link":"http:\/\/www.banbanli.net\/blog\/?p=2143","title":{"rendered":"[\u8f49\u8cbc]Google \u5927\u795e\u5728 Big Data \u61c9\u7528\u4e0a\u7684\u6280\u8853\u8207\u8ad6\u6587\u4ecb\u7d39\uff0c\u597d\u795e\uff01"},"content":{"rendered":"<p>http:\/\/techorange.com\/2013\/05\/14\/big-data-beyond-mapreduce\/<\/p>\n<p>&nbsp;<\/p>\n<p>\u76ee\u524d Big Data \u7684\u76f8\u95dc\u61c9\u7528\u6709\u4e0d\u5c11\u90fd\u662f\u5f9e MapReduce \u884d\u751f\u800c\u51fa\u7684\uff0c\u4f46\uff0c\u82e5\u628a\u7126\u9ede\u79fb\u5230\u5373\u6642\u8cc7\u6599\uff08Real-Time Data\uff09\u7684\u9700\u6c42\u4e0a\u6642\u5c31\u6703\u767c\u73fe\u5b83\u7684\u4e0d\u8db3\u4e4b\u8655\u3002<\/p>\n<p>\u56e0\u6b64\uff0c\u672c\u7bc7\u6587\u7ae0\u5c07\u8207\u5404\u4f4d\u8b80\u8005\u5206\u4eab\uff0c\u5f9e Google \u767c\u4f48 GFS\u3001Big Data \u8207 MapReduce \u9019\u4e9b\u6280\u8853\u958b\u59cb\uff0c\u5230\u9019\u4e9b\u6280\u8853\u767c\u5c55\u7684\u73fe\u6cc1\u8207\u5176\u6240\u906d\u9047\u5230\u7684\u74f6\u9838\uff0c\u4ee5\u53ca Google \u70ba\u4e86\u89e3\u6c7a\u9019\u4e9b\u554f\u984c\u63d0\u51fa\u4e86\u54ea\u4e9b\u56e0\u61c9\u7684\u6280\u8853\u3002<\/p>\n<ul>\n<li>MapReduce\uff0cGFS \u8207 Bigtable\uff0c\u5e36\u52d5\u4e86 Big Data \u61c9\u7528\u6280\u8853\u7684\u767c\u5c55<\/li>\n<\/ul>\n<p>Google \u5728 2003 \u5e74\u767c\u8868\u4e86\u7b2c\u4e00\u7bc7\u8ad6\u6587 \u3008<a title=\"The Google File System\" href=\"http:\/\/research.google.com\/archive\/gfs.html\" target=\"_blank\">The Google File System<\/a>\u3009\u3002 \u6587\u4e2d\u6558\u8ff0\uff0cGFS\uff08Google File System\uff09\u662f\u4e00\u500b\u5206\u6563\u5f0f\u6a94\u6848\u7cfb\u7d71\uff0c\u7531\u6578\u767e\u500b\u53e2\u96c6\uff08Cluster\uff09\u6240\u7d44\u6210\u3002\u7c21\u55ae\u4f86\u8aaa\uff0c\u5132\u5b58\u5728 GFS \u7684\u6a94\u6848\u6703\u88ab\u5207\u5272\u6210 \u00a064 MB \u5de6\u53f3\u7684\u8cc7\u6599\u584a\uff08Chunk\uff09\uff0c\u5176\u5229\u7528\u91cd\u8907\u7684\u65b9\u5f0f\uff08Redundant Fashion\uff09\u5132\u5b58\u5728\u53e2\u96c6\u4e2d\u3002<!--more--><\/p>\n<p>2004 \u5e74\uff0cGoogle \u767c\u8868\u4e86 MapReduce \u8ad6\u6587 \u3008<a title=\"MapReduce:&amp;nbsp;Simplified&amp;nbsp;Data&amp;nbsp;Processing&amp;nbsp;on&amp;nbsp;Large&amp;nbsp;Clusters\" href=\"http:\/\/research.google.com\/archive\/mapreduce.html\" target=\"_blank\">MapReduce:\u00a0Simplified\u00a0Data\u00a0Processing\u00a0on\u00a0Large\u00a0Clusters<\/a>\u3009\uff0c\u5982\u4eca\uff0cMapReuce \u53ef\u4ee5\u8aaa\u5e7e\u4e4e\u5df2\u7d93\u8ddf Big Data \u5283\u4e0a\u7b49\u865f\u4e86\u3002<\/p>\n<p>Google \u5229\u7528 MapReuce \u6f14\u7b97\u6cd5\u4f86\u8a08\u7b97\u67e5\u8a62\u7d22\u5f15\uff08Search Index\uff09\uff0c\u8b93\u4f7f\u7528\u8005\u80fd\u5728\u6700\u77ed\u7684\u6642\u9593\u5167\u5f9e Internet \u4e0a\u627e\u5230\u81ea\u5df1\u6240\u9700\u8981\/\u67e5\u8a62\u7684\u8cc7\u6599\u3002<\/p>\n<p>2006 \u5e74\uff0cGoogle \u767c\u8868\u4e86 Bigtable \u8ad6\u6587 \u3008<a title=\"Bigtable: A Distributed Storage System for Structured Data\" href=\"http:\/\/research.google.com\/archive\/bigtable.html\" target=\"_blank\">Bigtable: A Distributed Storage System for Structured Data<\/a>\u3009\uff0c\u800c Bigtable \u5e36\u9818\u4e86\u8a31\u591a NoSQL \u8cc7\u6599\u5eab\u7684\u6280\u8853\u61c9\u7528\u767c\u5c55\uff0c\u50cf\u662f Cassandra\u3001HBase \u7b49\u7b49\u3002<\/p>\n<p>\u5176\u4e2d\uff0cCassandra \u7684\u67b6\u69cb\u5c31\u6574\u5408\u4e86 Bigtable\uff08\u8cc7\u6599\u6a21\u578b\u3001SSTables \u53ca Write-Through-Logs\uff09\u8207 Amazon \u7684 Dynamo \u8cc7\u6599\u5eab\uff08Peer-to-Peer Clustering Model\uff09\u3002<\/p>\n<ul>\n<li>Percolator\uff0c\u80fd\u5920\u89e3\u6c7a MapReduce \u7121\u6cd5\u8655\u7406\u500b\u5225\u66f4\u65b0\u7684\u554f\u984c<\/li>\n<\/ul>\n<p>\u96a8\u8457 Internet \u7684\u7db2\u9801\u5448\u73fe\u6307\u6578\u589e\u52a0\uff0cMapReduce \u6bcf\u6b21\u90fd\u8981\u5168\u9762\u5730\u91cd\u65b0\u8a08\u7b97\u67e5\u8a62\u7d22\u5f15\u662f\u975e\u5e38\u4e0d\u5207\u5be6\u969b\u7684\u3002\u56e0\u6b64\uff0cGoogle \u70ba\u4e86\u63d0\u5347\u7cfb\u7d71\u7684\u6548\u80fd\uff0c\u958b\u767c\u4e86\u4e00\u500b\u66f4\u6709\u50f9\u503c\u7684\u5206\u6563\u5f0f\u8a08\u7b97\u7cfb\u7d71\uff1aPercolator\u3002<\/p>\n<p>Google \u5728 2010 \u5e74\u767c\u8868\u4e86\u76f8\u95dc\u7684\u8ad6\u6587 \u3008<a title=\"&amp;nbsp;Large-scale Incremental Processing Using Distributed Transactions and Notifications\" href=\"http:\/\/research.google.com\/pubs\/pub36726.html\" target=\"_blank\">\u00a0Large-scale Incremental Processing Using Distributed Transactions and Notifications<\/a>\u3009\uff0c\u6587\u4e2d\u6558\u8ff0 Google \u5982\u4f55\u5728\u7db2\u8def\u641c\u5c0b\u7d22\u5f15\uff08Web Search Index\uff09\u7684\u6280\u8853\u6301\u7e8c\u5730\u7dad\u6301\u7cbe\u9032\u3002\u4f8b\u5982\uff0cMapReduce \u505a\u8a08\u7b97\u6642\u7121\u6cd5\u8655\u7406\u5c40\u90e8\u7684\u66f4\u65b0\uff0c\u56e0\u6b64\uff0c\u5728\u6548\u80fd\u7684\u6539\u5584\u90e8\u4efd\u662f\u5f88\u6709\u9650\u7684\uff0c\u800c Percolator \u5247\u5f4c\u88dc\u4e86\u9019\u500b\u5f31\u9ede\u3002<\/p>\n<p>\u4f46\u662f\uff0c\u5404\u4f4d\u8b80\u8005\u5343\u842c\u4e0d\u8981\u8aa4\u4ee5\u70ba Percolator \u662f\u7528\u4f86\u53d6\u4ee3\u00a0 MapReduce \u7684\u3002<\/p>\n<p>Percolator \u662f\u5efa\u7acb\u65bc Bigtable \u4e4b\u4e0a\u7684\u61c9\u7528\uff0c\u5b83\u52a0\u5165\u4e86\u5c0d\u8868\uff08Table\uff09\u8207\u7d00\u9304\uff08Row\uff09\u7684\u4ea4\u6613\uff08Transaction\uff09\u8207\u9396\u5b9a\uff08Lock\uff09\u6a5f\u5236\uff0c\u4e5f\u5c31\u662f\u7576 GFS \u505a\u8868\u7684\u6383\u63cf\u6642\uff0c\u4e00\u65e6\u767c\u73fe\u6709\u66f4\u65b0\u904e\u7684\u7d00\u9304\uff0c\u5c31\u6703\u900f\u904e\u89f8\u767c\u7a0b\u5e8f\uff08Trigger\uff09\u544a\u77e5\u9019\u500b\u6539\u8b8a\uff0c\u518d\u4f9d\u64da\u8b80\u53d6\uff08Read\uff09\u6216\u5beb\u5165\uff08Write\uff09\u7684\u8acb\u6c42\uff0c\u5728\u4e0d\u540c\u968e \u6bb5\u7684\u5de5\u4f5c\u904e\u7a0b\u4e2d\uff0c\u91dd\u5c0d\u8cc7\u6599\u8868\u6216\u8a18\u9304\u505a\u9396\u5b9a\u6216\u91cb\u653e\u7684\u7ba1\u7406\u6a5f\u5236\u3002\u900f\u904e\u9019\u6a23\u5b50\u7684\u65b9\u5f0f\uff0c\u4f86\u5b8c\u6210\u5c40\u90e8\u500b\u9ad4\u7684\u66f4\u65b0\u3002<\/p>\n<ul>\n<li>Pregel\uff1a\u7528\u4f86\u8655\u7406\u7db2\u7d61\u793e\u4ea4\u95dc\u4fc2\u7684\u5716\u578b\u7d50\u69cb\u8a08\u7b97<\/li>\n<\/ul>\n<p>Google \u70ba\u4e86\u505a\u7db2\u8def\u793e\u4ea4\u95dc\u4fc2\u7684\u5716\u578b\u7d50\u69cb\u5206\u6790\uff0c\u958b\u59cb\u91dd\u5c0d\u5716\u578b\u7d50\u69cb\u63a2\u52d8\u505a\u76f8\u95dc\u7684\u7814\u7a76\u8207\u767c\u5c55\uff0c\u4e26\u5728 2010 \u5e74\u767c\u8868\u4e86\u76f8\u95dc\u7684\u8ad6\u6587 \u3008<a title=\"Pregel: A System for Large-Scale Graph Processing\" href=\"http:\/\/kowshik.github.io\/JPregel\/pregel_paper.pdf\" target=\"_blank\">Pregel: A System for Large-Scale Graph Processing<\/a>\u3009\u3002<\/p>\n<p>\u7531\u65bc\u91dd\u5c0d\u5927\u578b\u7684\u5716\u578b\u7d50\u69cb\u505a\u8655\u7406\u662f\u975e\u5e38\u8907\u96dc\u4e5f\u5177\u6709\u6311\u6230\u6027\u7684\uff0c\u5c24\u5176\u662f\u7db2\u8def\u7684\u5206\u6563\u5f0f\u8655\u7406\u8b93\u96e3\u5ea6\u53c8\u63d0\u9ad8\u4e86\u8a31\u591a\uff0c\u56e0\u6b64\uff0cPregel \u7684\u8a08\u7b97\u8981\u6bd4MapReduce \u7684\u8a08\u7b97\u8981\u8907\u96dc\u8a31\u591a\uff0c\u5176\u4e3b\u8981\u662f\u5229\u7528BSP\uff08Bluk Synchronous Parallell\uff09\u3001PageRank\u3001Bipartite Matching \u7b49\u6f14\u7b97\u6cd5\u4f86\u505a\u8a08\u7b97\u7684\u5be6\u8e10\u3002\u800c\u5728\u8ad6\u6587\u4e2d\u4f60\u4e5f\u53ef\u4ee5\u770b\u5230\u4e0a\u8ff0\u9019\u4e9b\u65b9\u6cd5\u7684\u5be6\u8e10\u3002<\/p>\n<ul>\n<li>Dremel\uff1a\u53ea\u8981\u82b1\u5e7e\u79d2\u9418\u6642\u9593\u5c31\u53ef\u4ee5\u5206\u6790 PB \u7b49\u7d1a\u7684\u6578\u64da<\/li>\n<\/ul>\n<p>\u5728 2010 \u5e74\uff0cGoogle \u9084\u540c\u6642\u767c\u8868\u4e86\u4e00\u7bc7\u95dc\u65bc <a title=\"Dremel\" href=\"http:\/\/research.google.com\/pubs\/pub36632.html\" target=\"_blank\">Dremel<\/a> \u7684\u8ad6\u6587\uff0c\u5167\u5bb9\u6558\u8ff0 Dremel \u662f\u4e00\u500b\u5229\u7528 SQL-like Language \u7684\u4e92\u52d5\u5f0f\u8cc7\u6599\u5eab\u7cfb\u7d71\uff0c\u7528\u4f86\u5132\u5b58\u7d50\u69cb\u5316\u8cc7\u6599\u3002<\/p>\n<p>Dremel \u7684\u7279\u8272\u662f\uff0c\u4ee5\u5217\u5132\u5b58\u70ba\u4e3b\uff0c\u4ee5\u6e1b\u5c11 CPU \u8207\u78c1\u789f\u7684\u8b80\u53d6\uff0c\u9032\u800c\u9054\u5230\u5feb\u901f\u8b80\u53d6\u5c40\u90e8\u8cc7\u6599\u7684\u76ee\u7684\uff1b\u5c07\u67e5\u8a62\u7684\u4efb\u52d9\u5207\u5272\u6210\u591a\u500b\u5c0f\u4efb\u52d9\uff0c\u4ee5\u9054\u5230\u5e73\u884c\u8655\u7406\u7684\u76ee\u7684\uff1b\u652f\u63f4 Nested \u6578\u64da\u6a21\u578b\uff0c\u4f46\u53ea\u63d0\u4f9b\u552f\u8b80\u529f\u80fd\u3002<\/p>\n<ul>\n<li>Big Data \u7684\u76f8\u95dc\u61c9\u7528\uff0c\u9700\u8981\u7684\u4e0d\u53ea\u662f MapReduce<\/li>\n<\/ul>\n<p>Google \u4e26\u6c92\u6709\u5728\u63d0\u51fa MapReduce \u4e4b\u5f8c\uff0c\u5c31\u505c\u6b62\u4e86\u67e5\u8a62\u6280\u8853\u7684\u767c\u5c55\uff0c\u4ed6\u5011\u4e0d\u65b7\u5730\u767c\u5c55\u65b0\u7684\u6280\u8853\u4ee5\u88dc\u5f37 MapReduce \u7684\u4e0d\u8db3\u4e4b\u8655\uff0c\u9019\u5c0d Big Data \u7684\u767c\u5c55\u662f\u6709\u76ca\u7684\u3002<\/p>\n<p>\u7562\u7adf\uff0cMapReduce \u4e0d\u662f\u842c\u80fd\u7684\uff0c\u4ee5\u76ee\u524d\u7684\u72c0\u6cc1\u4f86\u8aaa\uff0c\u9084\u662f\u6709\u8a31\u591a\u7684\u554f\u984c\u5c1a\u5f85\u89e3\u6c7a\uff0c\u4e0d\u904e\uff0cGoogle \u6240\u767c\u5c55\u51fa\u4f86\u7684\u6280\u8853\uff0c\u5e36\u9818\u4e86\u8a31\u591a\u5176\u4ed6\u81ea\u7531\u8edf\u9ad4\u7684\u767c\u5c55\uff0c\u50cf Apache Drill\u3001Apache Giraph \u4ee5\u53ca Stanford\u2019s GPS \u7b49\u7b49\uff0c\u9019\u6216\u8a31\u662f\u8eab\u70ba\u4f7f\u7528\u8005\u7684\u6211\u5011\u6700\u6a02\u65bc\u898b\u5230\u7684\u767c\u5c55\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>http:\/\/techorange.com\/2013\/05\/14\/big-data-beyond-mapreduce\/ &nbsp; \u76ee\u524d Big Data \u7684\u76f8\u95dc\u61c9\u7528\u6709\u4e0d\u5c11\u90fd\u662f\u5f9e MapReduce \u884d\u751f\u800c\u51fa\u7684\uff0c\u4f46\uff0c\u82e5\u628a\u7126\u9ede\u79fb\u5230\u5373\u6642\u8cc7\u6599\uff08Real-Time Data\uff09\u7684\u9700\u6c42\u4e0a\u6642\u5c31\u6703\u767c\u73fe\u5b83\u7684\u4e0d\u8db3\u4e4b\u8655\u3002 \u56e0\u6b64\uff0c\u672c\u7bc7\u6587\u7ae0\u5c07\u8207\u5404\u4f4d\u8b80\u8005\u5206\u4eab\uff0c\u5f9e Google \u767c\u4f48 GFS\u3001Big Data \u8207 MapReduce \u9019\u4e9b\u6280\u8853\u958b\u59cb\uff0c\u5230\u9019\u4e9b\u6280\u8853\u767c\u5c55\u7684\u73fe\u6cc1\u8207\u5176\u6240\u906d\u9047\u5230\u7684\u74f6\u9838\uff0c\u4ee5\u53ca Google \u70ba\u4e86\u89e3\u6c7a\u9019\u4e9b\u554f\u984c\u63d0\u51fa\u4e86\u54ea\u4e9b\u56e0\u61c9\u7684\u6280\u8853\u3002 MapReduce\uff0cGFS \u8207 Bigtable\uff0c\u5e36\u52d5\u4e86 Big Data \u61c9\u7528\u6280\u8853\u7684\u767c\u5c55 Google \u5728 2003 \u5e74\u767c\u8868\u4e86\u7b2c\u4e00\u7bc7\u8ad6\u6587 \u3008The Google File System\u3009\u3002 \u6587\u4e2d\u6558\u8ff0\uff0cGFS\uff08Google File System\uff09\u662f\u4e00\u500b\u5206\u6563\u5f0f\u6a94\u6848\u7cfb\u7d71\uff0c\u7531\u6578\u767e\u500b\u53e2\u96c6\uff08Cluster\uff09\u6240\u7d44\u6210\u3002\u7c21\u55ae\u4f86\u8aaa\uff0c\u5132\u5b58\u5728 GFS \u7684\u6a94\u6848\u6703\u88ab\u5207\u5272\u6210 \u00a064 MB \u5de6\u53f3\u7684\u8cc7\u6599\u584a\uff08Chunk\uff09\uff0c\u5176\u5229\u7528\u91cd\u8907\u7684\u65b9\u5f0f\uff08Redundant Fashion\uff09\u5132\u5b58\u5728\u53e2\u96c6\u4e2d\u3002<\/p>\n","protected":false},"author":3,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[51],"tags":[],"class_list":["post-2143","post","type-post","status-publish","format-standard","hentry","category-big-data"],"_links":{"self":[{"href":"http:\/\/www.banbanli.net\/blog\/index.php?rest_route=\/wp\/v2\/posts\/2143","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/www.banbanli.net\/blog\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.banbanli.net\/blog\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.banbanli.net\/blog\/index.php?rest_route=\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"http:\/\/www.banbanli.net\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=2143"}],"version-history":[{"count":1,"href":"http:\/\/www.banbanli.net\/blog\/index.php?rest_route=\/wp\/v2\/posts\/2143\/revisions"}],"predecessor-version":[{"id":2144,"href":"http:\/\/www.banbanli.net\/blog\/index.php?rest_route=\/wp\/v2\/posts\/2143\/revisions\/2144"}],"wp:attachment":[{"href":"http:\/\/www.banbanli.net\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2143"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.banbanli.net\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2143"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.banbanli.net\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2143"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}