{"id":1284,"date":"2013-12-09T15:00:05","date_gmt":"2013-12-09T06:00:05","guid":{"rendered":"http:\/\/kazu.tv\/blog\/?p=1284"},"modified":"2013-12-09T15:03:01","modified_gmt":"2013-12-09T06:03:01","slug":"dynamodb-s3-emr-cohort-analysis-3","status":"publish","type":"post","link":"https:\/\/kazu.tv\/blog\/2013\/12\/09\/dynamodb-s3-emr-cohort-analysis-3\/","title":{"rendered":"DynamoDB + S3 + EMR\u3067\u30b3\u30db\u30fc\u30c8\u5206\u6790(cohort analysis)\u3092\u3059\u308b(3)"},"content":{"rendered":"<p><a title=\"DynamoDB + S3 + EMR\u3067\u30b3\u30db\u30fc\u30c8\u5206\u6790(cohort analysis)\u3092\u3059\u308b(2)\" href=\"http:\/\/kazu.tv\/blog\/2013\/10\/09\/dynamodb-s3-emr-cohort-analysis-2\/\">\u524d\u56de<\/a>\u304b\u3089\u304b\u306a\u308a\u9593\u304c\u7a7a\u3044\u3066\u3057\u307e\u3063\u305f\u3051\u3069\u3001\u4eca\u56de\u3067\u5b8c\u7d50\u4e88\u5b9a\u3002<\/p>\n<p>\u524d\u56de\u306fHive\u306e\u8a71\u3092\u4e2d\u5fc3\u306b\u3001S3\u306b\u7f6e\u3044\u305f\u30d5\u30a1\u30a4\u30eb\u3092Hive\u3067\u3069\u3046\u6271\u3046\u304b\u306a\u3069\u306b\u3064\u3044\u3066\u66f8\u3044\u305f\u3002\u307e\u305f\u3001<a title=\"DynamoDB + S3 + EMR\u3067\u30b3\u30db\u30fc\u30c8\u5206\u6790(cohort analysis)\u3092\u3059\u308b(2)\" href=\"http:\/\/kazu.tv\/blog\/2013\/10\/09\/dynamodb-s3-emr-cohort-analysis-2\/\">\u7b2c1\u56de<\/a>\u3067\u306f\u5168\u4f53\u306e\u6d41\u308c\u3092\u66f8\u3044\u305f\u306e\u3067\u3001\u3069\u3093\u306a\u3053\u3068\u3092\u3084\u308b\u304b\u306f\u8a73\u3057\u304f\u306f\u305d\u3061\u3089\u3092\u53c2\u7167\u3002<\/p>\n<p>\u4eca\u56de\u306f\u3001DynamoDB\u306b\u66f8\u304d\u8fbc\u3093\u3067\u3044\u308b\u30c7\u30fc\u30bf\u3092\u5b9a\u671f\u7684\u306bS3\u306b\u30a8\u30af\u30b9\u30dd\u30fc\u30c8\u3057\u305f\u308a\u3001MySQL\u304b\u3089\u30a8\u30af\u30b9\u30dd\u30fc\u30c8\u3057\u305f\u30c7\u30fc\u30bf\u306b\u5bfe\u3057\u3066\u3001EMR\u4e0a\u306eHive\u304b\u3089\u30af\u30a8\u30ea\u30fc\u3092\u5b9f\u884c\u3057\u3066\u7d50\u679c\u3092\u53d6\u5f97\u3057\u3066\u307f\u308b\u3002<\/p>\n<p><!--more--><\/p>\n<h2>DynamoDB\u304b\u3089S3\u306b\u30a8\u30af\u30b9\u30dd\u30fc\u30c8<\/h2>\n<p>DynamoDB\u306f\u3001\u4e8b\u524d\u306b\u8a2d\u5b9a\u3057\u305f\u8aad\u307f\u66f8\u304d\u306e\u30b9\u30eb\u30fc\u30d7\u30c3\u30c8\u4ee5\u4e0a\u306e\u30a2\u30af\u30bb\u30b9\u306f\u51fa\u6765\u306a\u3044\u3057\u3001\u540c\u3058\u30c7\u30fc\u30bf\u306b\u5bfe\u3057\u3066\u9055\u3046\u30af\u30a8\u30ea\u30fc\u3092\u8907\u6570\u5b9f\u884c\u3059\u308b\u5834\u5408\u3068\u304b\u306b\u7121\u99c4\u306a\u306e\u3067\u3001\u904e\u53bb\u30c7\u30fc\u30bf\u3092\u5b9a\u671f\u7684\u306bS3\u306b\u30a8\u30af\u30b9\u30dd\u30fc\u30c8\u3059\u308b\u4e8b\u306b\u3057\u305f\u3002<\/p>\n<h3>\u30c6\u30fc\u30d6\u30eb\u5b9a\u7fa9<\/h3>\n<p>\u307e\u305a\u306fDynamoDB\u306e\u30c6\u30fc\u30d6\u30eb\u3092Hive\u306b\u8a8d\u8b58\u3055\u305b\u308b\u305f\u3081\u306b\u3001\u4ee5\u4e0b\u306e\u3088\u3046\u306a\u30c6\u30fc\u30d6\u30eb\u3092\u4f5c\u6210\u3059\u308b\u3002<\/p>\n<pre class=\"brush: sql;\">CREATE EXTERNAL TABLE dynamo_log_users (\r\nuser_id string,\r\naction string,\r\ndevice string,\r\ntime string,\r\ncreatetime bigint\r\n)\r\nSTORED BY\r\n'org.apache.hadoop.hive.dynamodb.DynamoDBStorageHandler' \r\nTBLPROPERTIES (\r\n\"dynamodb.table.name\" = \"log_users\",\r\n\"dynamodb.column.mapping\" = \r\n\"action:action,time:time,user_id:user_id,createtime:createtime,device:device\");<\/pre>\n<p>\u6b21\u306b\u3001\u4e0a\u306e\u30c6\u30fc\u30d6\u30eb\uff08\u5b9f\u4f53\u306fDynamoDB\uff09\u304b\u3089\u30a8\u30af\u30b9\u30dd\u30fc\u30c8\u3059\u308b\u30c7\u30fc\u30bf\u3092\u683c\u7d0d\u3059\u308b\u30c6\u30fc\u30d6\u30eb\uff08\u5b9f\u4f53\u306fS3\u4e0a\u306e\u30d5\u30a1\u30a4\u30eb\uff09\u3092\u4f5c\u6210\u3002\u4e0b\u306e\u4f8b\u3067\u306f\u3001createtime\u30d5\u30a3\u30fc\u30eb\u30c9\u306e\u5024\u306b\u3088\u3063\u3066\u3001year\u3068month\u3067\u30d1\u30fc\u30c6\u30a3\u30b7\u30e7\u30f3\u5206\u5272\u3059\u308b\u3088\u3046\u306a\u8a2d\u5b9a\u306b\u3057\u3066\u3044\u308b\u304c\u3001\u30c7\u30fc\u30bf\u91cf\u7b49\u3001\u305d\u306e\u4ed6\u306e\u5236\u7d04\u306b\u5fdc\u3058\u3066\u30011\u65e5\u5358\u4f4d\u3067\u306e\u30d1\u30fc\u30c6\u30a3\u30b7\u30e7\u30f3\u4f5c\u6210\u3067\u3082\u826f\u3044\u304b\u3082\u3002<\/p>\n<pre class=\"brush: sql;\">CREATE EXTERNAL TABLE log_users (\r\nuser_id string,\r\naction string,\r\ndevice string,\r\ntime string,\r\ncreatetime bigint\r\n)\r\nPARTITIONED BY (year string, month string)\r\nROW FORMAT DELIMITED\r\n\u00a0\u00a0 FIELDS TERMINATED BY '\\t'\r\n\u00a0\u00a0 LINES TERMINATED BY '\\n'\r\nLOCATION 's3:\/\/bucket\/hive\/log_users\/';<\/pre>\n<h3>Dynamic Partitions<\/h3>\n<p>Hive\u4e0a\u306b\u30c6\u30fc\u30d6\u30eb\u304c\u51fa\u6765\u3061\u3083\u3048\u3070\u3001\u5f8c\u306f\u901a\u5e38\u901a\u308aINSERT OVERWRITE TABLE\u3067\u30c7\u30fc\u30bf\u3092\u30b3\u30d4\u30fc\u3059\u308b\u3060\u3051\u3001\u306a\u3093\u3060\u3051\u3069\u3001\u305d\u306e\u524d\u306b\u4ee5\u4e0b\u306e\u3088\u3046\u306a\u8a2d\u5b9a\u3092\u3057\u3066\u304a\u304f\u5fc5\u8981\u304c\u3042\u308b\u3002\u8a73\u3057\u304f\u306fWiki\u306e<a title=\"Dynamic Partitions\" href=\"https:\/\/cwiki.apache.org\/confluence\/display\/Hive\/DynamicPartitions\" target=\"_blank\">Dynamic Partitions\u306e\u9805<\/a>\u3092\u53c2\u7167\u3002<\/p>\n<pre class=\"brush: sql;\">SET hive.exec.dynamic.partition=true;\r\nSET hive.exec.dynamic.partition.mode=nonstrict;<\/pre>\n<h3>\u30a8\u30af\u30b9\u30dd\u30fc\u30c8<\/h3>\n<p>\u305d\u306e\u5f8c\u3001\u901a\u5e38\u901a\u308aINSERT OVERWRITE TABLE\u3067DynamoDB\u306e\u30c6\u30fc\u30d6\u30eb\uff08dynamo_log_users\uff09\u304b\u3089S3\u306e\u30c6\u30fc\u30d6\u30eb\uff08log_users\uff09\u306b\u30c7\u30fc\u30bf\u3092\u30b3\u30d4\u30fc\u3059\u308b\u305f\u3081\u3001\u4ee5\u4e0b\u306e\u3088\u3046\u306a\u30af\u30a8\u30ea\u30fc\u3092\u5b9f\u884c\u3002<\/p>\n<pre class=\"brush: sql;\">INSERT OVERWRITE TABLE log_users\r\nPARTITION(year, month)\r\nSELECT *,\r\n       from_unixtime(createtime, 'yyyy') year,\r\n       from_unixtime(createtime, 'MM') month\r\n FROM dynamo_log_users\r\nWHERE from_unixtime(createtime, 'yyyyMM') &gt;= '201311'\r\n  AND from_unixtime(createtime, 'yyyyMM') &lt;  '201312'\r\nDISTRIBUTE BY year, month;<\/pre>\n<h2>MySQL\u306e\u30c6\u30fc\u30d6\u30eb\u3092\u30a8\u30af\u30b9\u30dd\u30fc\u30c8<\/h2>\n<p>MySQL\u306b\u683c\u7d0d\u3055\u308c\u3066\u3044\u308b\u30e6\u30fc\u30b6\u30fc\u60c5\u5831\u3082\u5b9a\u671f\u7684\u306bS3\u306b\u30a8\u30af\u30b9\u30dd\u30fc\u30c8\u3059\u308b\u3002CSV\u304bTSV\u3067\u30a8\u30af\u30b9\u30dd\u30fc\u30c8\u3059\u308b\u306e\u304c\u826f\u3044\u3068\u601d\u308f\u308c\u308b\u3002\u3061\u3087\u3063\u3068\u8abf\u3079\u305f\u611f\u3058\u3001\u4e00\u756a\u697d\u306a\u306e\u306f\u4ee5\u4e0b\u306e\u65b9\u6cd5\u3067TSV\u306b\u3059\u308b\u65b9\u6cd5\u3002<\/p>\n<pre class=\"brush: shell;\">echo \"SELECT * FROM users WHERE ...\" | \r\nmysql -h host -u user -ppass dbname --skip-column-names &gt; file<\/pre>\n<p>\u3042\u3068\u306f\u51fa\u6765\u305f\u30d5\u30a1\u30a4\u30eb\u3092s3cmd\u7b49\u3067S3\u306b\u30a2\u30c3\u30d7\u30ed\u30fc\u30c9\u3059\u308b\uff08\u8a73\u7d30\u306f\u7701\u7565\uff09\u3002<\/p>\n<h2>\u7d50\u679c\u7528\u306e\u30c6\u30fc\u30d6\u30eb\u4f5c\u6210<\/h2>\n<p>\u4ee5\u4e0b\u306e\u3088\u3046\u306b\u7d50\u679c\u7528\u30c6\u30fc\u30d6\u30eb\u3092\u4f5c\u6210\u3002<\/p>\n<pre class=\"brush: sql;\">CREATE EXTERNAL TABLE results_weekly\r\n(cohort_start_date string, value float)\r\nPARTITIONED BY (result_date string)\r\nROW FORMAT DELIMITED\r\n\u00a0\u00a0 FIELDS TERMINATED BY '\\t'\r\n\u00a0\u00a0 LINES TERMINATED BY '\\n'\r\nLOCATION 's3:\/\/bucket\/hive\/results_weekly\/';<\/pre>\n<p>\u30ab\u30e9\u30e0\u306f\u4ee5\u4e0b\u306e3\u3064\u3002<\/p>\n<ul>\n<li>\u30b3\u30db\u30fc\u30c8\u540d\uff08\u65e5\u4ed8\uff09\uff1a\u4f8b\u21922013\/11\/4\u9031<\/li>\n<li>\u30c7\u30fc\u30bf\u96c6\u8a08\u5bfe\u8c61\uff1a\u4f8b\u21922013\/12\/9\u9031<\/li>\n<li>\u30c7\u30fc\u30bf\uff1a\u4f8b\u219240.5<\/li>\n<\/ul>\n<h2>\u96c6\u8a08\u30af\u30a8\u30ea\u30fc\u5b9f\u884c<\/h2>\n<p>\u3053\u3053\u307e\u3067\u3067\u3088\u3046\u3084\u304f\u6e96\u5099\u304c\u6574\u3063\u305f\u306e\u3067\u3001\u63c3\u3063\u305f\u30c7\u30fc\u30bf\u306b\u5bfe\u3057\u3066Hive\u304b\u3089\u30af\u30a8\u30ea\u30fc\u3092\u5b9f\u884c\u3059\u308b\u3002<\/p>\n<p>\u4f8b\u3048\u3070\u30a2\u30af\u30c6\u30a3\u30d6\u30e6\u30fc\u30b6\u30fc\u6570\u3092\u6e2c\u308a\u305f\u3044\u5834\u5408\u3001\u8a72\u5f53\u671f\u9593\u4e2d\u306elog_users\u306buser_id\u304c\u542b\u307e\u308c\u308c\u3070\u30a2\u30af\u30c6\u30a3\u30d6\u3068\u3059\u308b\u3068\u3001\u4ee5\u4e0b\u306e\u3088\u3046\u306a\u30af\u30a8\u30ea\u30fc\u3068\u306a\u308b\u3002<\/p>\n<pre class=\"brush: sql;\">INSERT OVERWRITE TABLE results_weekly\r\nPARTITION(result_date)\r\nSELECT ucm.cohort, COUNT(DISTINCT(l.user_id)), dcm.cohort_week\r\n\u00a0 FROM user_cohort_map ucm\r\n\u00a0 JOIN log_users l ON ucm.user_id = l.user_id\r\n\u00a0 JOIN date_cohort_map dcm ON dcm.date = from_unixtime(l.createtime, 'yyyy-MM-dd')\r\nWHERE dcm.cohort_week = '2013-12-09' -- \u96c6\u8a08\u65e5\r\nGROUP BY ucm.cohort, dcm.cohort_week ;<\/pre>\n<p>\u5b9f\u969b\u306e\u696d\u52d9\u306b\u5408\u308f\u305b\u3066\u3001\u30c6\u30fc\u30d6\u30eb\u69cb\u9020\u3084\u30af\u30a8\u30ea\u30fc\u3068\u304b\u3092\u5909\u3048\u3066\u307f\u3066\u4e0b\u3055\u3044\u3002<\/p>\n<h2>\u307e\u3068\u3081<\/h2>\n<p>\u5206\u304b\u3063\u3066\u3057\u307e\u3048\u3070\u3001AWS\u3067\u30b3\u30db\u30fc\u30c8\u5206\u6790\u3092\u3059\u308b\u306e\u306f\u305d\u3093\u306a\u306b\u5927\u5909\u3067\u306f\u306a\u3044\u3002Elastic MapReduce, S3, DynamoDB\u306a\u3069\u306e\u4f7f\u3048\u308b\u30b5\u30fc\u30d3\u30b9\u304c\u6700\u521d\u304b\u3089\u63c3\u3063\u3066\u3044\u308b\u3068\u3044\u3046\u306e\u304c\u5927\u304d\u3044\u3002<\/p>\n<p>\u3068\u8a00\u3063\u3066\u3082\u30c7\u30fc\u30bf\u304c\u5897\u3048\u3066\u304f\u308b\u3068\u8272\u3005\u82e6\u52b4\u304c\u3068\u601d\u3046\u306e\u3067\u3001\u30c6\u30fc\u30d6\u30eb\u8a2d\u8a08\u306a\u3069\u306f\u3057\u3063\u304b\u308a\u8003\u3048\u3066\u5b9f\u884c\u3059\u308b\u4e8b\u3092\u30aa\u30b9\u30b9\u30e1\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u524d\u56de\u304b\u3089\u304b\u306a\u308a\u9593\u304c\u7a7a\u3044\u3066\u3057\u307e\u3063\u305f\u3051\u3069\u3001\u4eca\u56de\u3067\u5b8c\u7d50\u4e88\u5b9a\u3002 \u524d\u56de\u306fHive\u306e\u8a71\u3092\u4e2d\u5fc3\u306b\u3001S3\u306b\u7f6e\u3044\u305f\u30d5\u30a1\u30a4\u30eb\u3092Hive\u3067\u3069\u3046\u6271\u3046\u304b\u306a\u3069\u306b\u3064\u3044\u3066\u66f8\u3044\u305f\u3002\u307e\u305f\u3001\u7b2c1\u56de\u3067\u306f\u5168\u4f53\u306e\u6d41\u308c\u3092\u66f8\u3044\u305f\u306e\u3067\u3001\u3069\u3093\u306a\u3053\u3068\u3092\u3084\u308b\u304b\u306f\u8a73\u3057\u304f\u306f\u305d\u3061\u3089&hellip;<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_monsterinsights_skip_tracking":false,"_monsterinsights_sitenote_active":false,"_monsterinsights_sitenote_note":"","_monsterinsights_sitenote_category":0,"footnotes":""},"categories":[15],"tags":[913,911,914,912],"class_list":["post-1284","post","type-post","status-publish","format-standard","hentry","category-15","tag-dynamodb","tag-emr","tag-hive","tag-s3"],"_links":{"self":[{"href":"https:\/\/kazu.tv\/blog\/wp-json\/wp\/v2\/posts\/1284","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/kazu.tv\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/kazu.tv\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/kazu.tv\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/kazu.tv\/blog\/wp-json\/wp\/v2\/comments?post=1284"}],"version-history":[{"count":4,"href":"https:\/\/kazu.tv\/blog\/wp-json\/wp\/v2\/posts\/1284\/revisions"}],"predecessor-version":[{"id":1288,"href":"https:\/\/kazu.tv\/blog\/wp-json\/wp\/v2\/posts\/1284\/revisions\/1288"}],"wp:attachment":[{"href":"https:\/\/kazu.tv\/blog\/wp-json\/wp\/v2\/media?parent=1284"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/kazu.tv\/blog\/wp-json\/wp\/v2\/categories?post=1284"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/kazu.tv\/blog\/wp-json\/wp\/v2\/tags?post=1284"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}