{"id":142681,"date":"2020-08-04T15:01:57","date_gmt":"2020-08-04T07:01:57","guid":{"rendered":"http:\/\/4563.org\/?p=142681"},"modified":"2020-08-04T15:01:57","modified_gmt":"2020-08-04T07:01:57","slug":"spark%e7%9a%84%e7%ae%97%e5%ad%90transformation%e5%92%8caction","status":"publish","type":"post","link":"http:\/\/4563.org\/?p=142681","title":{"rendered":"Spark\u7684\u7b97\u5b50Transformation\u548cAction"},"content":{"rendered":"

<\/a>RDD\u7684\u7b97\u5b50\u5206\u7c7b<\/h3>\n

<\/a>transformation\uff08\u8f6c\u6362\uff09<\/h4>\n

\u6839\u636e\u5df2\u7ecf\u5b58\u5728\u7684rdd\u8f6c\u6362\u751f\u6210\u4e00\u4e2a\u65b0\u7684rdd, \u5b83\u662f\u5ef6\u8fdf\u52a0\u8f7d\uff0c\u5b83\u4e0d\u4f1a\u7acb\u5373\u6267\u884c<\/p>\n

<\/a>action (\u52a8\u4f5c)<\/h4>\n

\u5b83\u4f1a\u771f\u6b63\u89e6\u53d1\u4efb\u52a1\u7684\u8fd0\u884c\uff0c\u5c06rdd\u7684\u8ba1\u7b97\u7684\u7ed3\u679c\u6570\u636e\u8fd4\u56de\u7ed9Driver\u7aef\uff0c\u6216\u8005\u662f\u4fdd\u5b58\u7ed3\u679c\u6570\u636e\u5230\u5916\u90e8\u5b58\u50a8\u4ecb\u8d28\u4e2d<\/p>\n

<\/a>RDD transformation<\/h3>\n

transformation API \u662f\u60f0\u6027\u7684\uff0c\u8c03\u7528\u8fd9\u4e9bAPI\u6bd4\u4e0d\u4f1a\u89e6\u53d1\u5b9e\u9645\u7684\u5206\u5e03\u5f0f\u6570\u636e\u8ba1\u7b97\uff0c\u800c\u4ec5\u4ec5\u662f\u5c06\u76f8\u5173\u4fe1\u606f\u8bb0\u5f55\u4e0b\u6765\uff0c\u76f4\u5230action API\u624d\u4f1a\u5f00\u59cb\u6570\u636e\u8ba1\u7b97\u3002<\/p>\n

Spark \u63d0\u4f9b\u4e86\u5927\u91cf\u7684 transformation API\uff0c\u4e0b\u9762\u5217\u4e3e\u4e86\u4e00\u4e9b\u5e38\u7528\u7684API:<\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
API<\/th>\n\u529f\u80fd<\/th>\n<\/tr>\n<\/thead>\n
map(func)<\/td>\n\u5c06 RDD \u4e2d\u7684\u5143\u7d20\uff0c\u901a\u8fc7 func \u51fd\u6570\u9010\u4e00\u6620\u5c04\u6210\u53e6\u5916\u4e00\u4e2a\u503c\uff0c\u5f62\u6210\u4e00\u4e2a\u65b0\u7684 RDD<\/td>\n<\/tr>\n
filter(func)<\/td>\n\u5c06 RDD \u4e2d\u4f7f\u7528 func \u51fd\u6570\u8fd4\u56de true \u7684\u5143\u7d20\u8fc7\u6ee4\u51fa\u6765\uff0c\u5f62\u6210\u4e00\u4e2a\u65b0\u7684 RDD<\/td>\n<\/tr>\n
flatMap(func<\/td>\n\u7c7b\u4f3c\u4e8emap\uff0c\u4f46\u6bcf\u4e00\u4e2a\u8f93\u5165\u5143\u7d20\u53ef\u4ee5\u88ab\u6620\u5c04\u4e3a0\u6216\u591a\u4e2a\u8f93\u51fa\u5143\u7d20\uff08\u6240\u4ee5func\u5e94\u8be5\u8fd4\u56de\u4e00\u4e2a\u5e8f\u5217\uff0c\u800c\u4e0d\u662f\u5355\u4e00\u5143\u7d20\uff09<\/td>\n<\/tr>\n
mapPartitions(func)<\/td>\n\u7c7b\u4f3c\u4e8e map\uff0c\u4f46\u72ec\u7acb\u5730\u5728 RDD \u7684\u6bcf\u4e00\u4e2a\u5206\u7247\u4e0a\u8fd0\u884c\uff0c\u56e0\u6b64\u5728\u7c7b\u578b\u4e3aT\u7684 RDD \u4e0a\u8fd0\u884c\u65f6\uff0cfunc\u7684\u51fd\u6570\u7c7b\u578b\u5fc5\u987b\u662fIterator[T] => Iterator[U]<\/td>\n<\/tr>\n
sample(withReplacement, fraction, seed)<\/td>\n\u6570\u636e\u91c7\u6837\u51fd\u6570\u3002\u6839\u636efraction\u6307\u5b9a\u7684\u6bd4\u4f8b\u5bf9\u6570\u636e\u8fdb\u884c\u91c7\u6837\uff0c\u53ef\u4ee5\u9009\u62e9\u662f\u5426\u4f7f\u7528\u968f\u673a\u6570\u8fdb\u884c\u66ff\u6362\uff0cseed \u7528\u4e8e\u6307\u5b9a\u968f\u673a\u6570\u751f\u6210\u5668\u79cd\u5b50<\/td>\n<\/tr>\n
union(otherDataset)<\/td>\n\u6c42\u4e24\u4e2a RDD (\u76ee\u6807 RDD \u4e0e\u6307\u5b9a RDD)\u7684\u5e76\u96c6\uff0c\u5e76\u4ee5 RDD \u5f62\u5f0f\u8fd4\u56de<\/td>\n<\/tr>\n
intersection(otherDataset)<\/td>\n\u6c42\u4e24\u4e2a RDD (\u76ee\u6807 RDD \u4e0e\u6307\u5b9a RDD )\u7684\u4ea4\u96c6\uff0c\u5e76\u4ee5 RDD \u5f62\u5f0f\u8fd4\u56de<\/td>\n<\/tr>\n
distinct([numTasks]))<\/td>\n\u5bf9\u76ee\u6807 RDD \u8fdb\u884c\u53bb\u91cd\u540e\u8fd4\u56de\u4e00\u4e2a\u65b0\u7684 RDD<\/td>\n<\/tr>\n
groupByKey([numTasks])<\/td>\n\u9488\u5bf9 key\/value \u7c7b\u578b\u7684 RDD\uff0c\u5c06 key \u76f8\u540c\u7684 value \u805a\u96c6\u5728\u4e00\u8d77\u3002\u9ed8\u8ba4\u4efb\u52a1\u5e76\u53d1\u5ea6\u4e0e\u7236 RDD \u76f8\u540c\uff0c\u53ef\u663e\u793a\u8bbe\u7f6e [numTasks]\u5927\u5c0f<\/td>\n<\/tr>\n
reduceByKey(func, [numTasks])<\/td>\n\u9488\u5bf9 key\/value \u7c7b\u578b\u7684 RDD\uff0c\u5c06 key \u76f8\u540c\u7684 value \u805a\u96c6\u5728\u4e00\u8d77\uff0c\u5c06\u5bf9\u6bcf\u7ec4value\uff0c\u6309\u7167\u51fd\u6570 func \u89c4\u7ea6\uff0c\u4ea7\u751f\u65b0\u7684 RDD<\/td>\n<\/tr>\n
aggregateByKey(zeroValue)(seqOp, combOp, [numTasks])<\/td>\n\u4e0e reduceByKey \u7c7b\u4f3c\uff0c\u4f46\u76ee\u6807 key\/value \u7684\u7c7b\u578b\u4e0e\u6700\u7ec8\u4ea7\u751f\u7684 RDD \u53ef\u80fd\u4e0d\u540c<\/td>\n<\/tr>\n
sortByKey([ascending], [numTasks])<\/td>\n\u9488\u5bf9 key\/value \u7c7b\u578b\u7684 RDD\uff0c\u6309\u7167 key \u8fdb\u884c\u6392\u5e8f\uff0c\u82e5 ascending \u4e3a true\uff0c\u5219\u4e3a\u5347\u5e8f\uff0c\u53cd\u4e4b\u4e3a\u964d\u5e8f<\/td>\n<\/tr>\n
join(otherDataset, [numTasks])<\/td>\n\u9488\u5bf9 key\/value \u7c7b\u578b\u7684 RDD\uff0c\u5bf9 (K,V) \u7c7b\u578b\u7684 RDD \u548c(K,W)\u7c7b\u578b\u7684RDD\u4e0a\u8c03\u7528\uff0c\u6309\u7167 key \u8fdb\u884c\u7b49\u503c\u8fde\u63a5\uff0c\u8fd4\u56de\u4e00\u4e2a\u76f8\u540ckey\u5bf9\u5e94\u7684\u6240\u6709\u5143\u7d20\u5728\u4e00\u8d77\u7684(K,(V,W))\u7684RDD \u76f8\u5f53\u4e8e\u5185\u8fde\u63a5\uff08\u6c42\u4ea4\u96c6\uff09<\/td>\n<\/tr>\n
cogroup(otherDataset, [numTasks])<\/td>\n\u5206\u7ec4\u51fd\u6570\uff0c\u5bf9(K,V)\u7c7b\u578b\u7684RDD\u548c(K,W)\u7c7b\u578b\u7684RD\u6309\u7167key\u8fdb\u884c\u5206\u7ec4\uff0c\u4ea7\u751f\u65b0\u7684 (K,(Iterable,Iterable)) \u7c7b\u578b\u7684RDD<\/td>\n<\/tr>\n
cartesian(otherDataset)<\/td>\n\u6c42\u4e24\u4e2a RDD \u7684\u7b1b\u5361\u5c14\u79ef<\/td>\n<\/tr>\n
coalesce(numPartitions)<\/td>\n\u91cd\u65b0\u5206\u533a, \u7f29\u51cf\u5206\u533a\u6570\uff0c\u7528\u4e8e\u5927\u6570\u636e\u96c6\u8fc7\u6ee4\u540e\uff0c\u63d0\u9ad8\u5c0f\u6570\u636e\u96c6\u7684\u6267\u884c\u6548\u7387<\/td>\n<\/tr>\n
repartition(numPartitions)<\/td>\n\u91cd\u65b0\u5206\u533a\uff0c\u5c06\u76ee\u6807 RDD \u7684 partition \u6570\u91cf\u91cd\u65b0\u8c03\u6574\u4e3a numPartitions\uff0c \u5c11\u53d8\u591a<\/td>\n<\/tr>\n
glom()<\/td>\n\u5c06RDD\u4e2d\u6bcf\u4e2apartition\u4e2d\u5143\u7d20\u8f6c\u6362\u4e3a\u6570\u7ec4,\u5e76\u751f \u6210\u65b0\u7684rdd2<\/td>\n<\/tr>\n
mapValues()<\/td>\n\u9488\u5bf9\u4e8e(K,V)\u5f62\u5f0f\u7684\u7c7b\u578b\u53ea\u5bf9V\u8fdb\u884c\u64cd\u4f5c<\/td>\n<\/tr>\n
cache<\/td>\nRDD\u7f13\u5b58\uff0c\u53ef\u4ee5\u907f\u514d\u91cd\u590d\u8ba1\u7b97\u4ece\u800c\u51cf\u5c11\u65f6\u95f4\uff0ccache \u5185\u90e8\u8c03\u7528\u4e86 persist \u7b97\u5b50\uff0ccache \u9ed8\u8ba4\u5c31\u4e00\u4e2a\u7f13\u5b58\u7ea7\u522b MEMORY-ONLY<\/td>\n<\/tr>\n
persist<\/td>\npersist \u53ef\u4ee5\u9009\u62e9\u7f13\u5b58\u7ea7\u522b<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

<\/a>RDD action<\/h3>\n

transformation \u7b97\u5b50\u5177\u6709\u60f0\u6027\u6267\u884c\u7684\u7279\u6027\uff0c\u4ed6\u4ec5\u4ec5\u662f\u8bb0\u5f55\u4e00\u4e9b\u539f\u4fe1\u606f\uff0c\u77e5\u9053\u9047\u5230action\u7b97\u5b50\u624d\u4f1a\u89e6\u53d1\u76f8\u5173transformation \u7b97\u5b50\u7684\u6267\u884c\uff0c<\/p>\n

Spark \u63d0\u4f9b\u4e86\u5927\u91cf\u7684 action API\uff0c\u4e0b\u9762\u5217\u4e3e\u4e86\u4e00\u4e9b\u5e38\u7528\u7684API:<\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
API<\/th>\n\u529f\u80fd<\/th>\n<\/tr>\n<\/thead>\n
reduce(func)<\/td>\n\u5c06RDD\u4e2d\u5143\u7d20\u524d\u4e24\u4e2a\u4f20\u7ed9\u8f93\u5165\u51fd\u6570\uff0c\u4ea7\u751f\u4e00\u4e2a\u65b0\u7684return\u503c\uff0c\u65b0\u4ea7\u751f\u7684return\u503c\u4e0eRDD\u4e2d\u4e0b\u4e00\u4e2a\u5143\u7d20\uff08\u7b2c\u4e09\u4e2a\u5143\u7d20\uff09\u7ec4\u6210\u4e24\u4e2a\u5143\u7d20\uff0c\u518d\u88ab\u4f20\u7ed9\u8f93\u5165\u51fd\u6570\uff0c\u76f4\u5230\u6700\u540e\u53ea\u6709\u4e00\u4e2a\u503c\u4e3a\u6b62<\/td>\n<\/tr>\n
collect()<\/td>\n\u5c06 RDD \u4ee5\u6570\u7ec4\u7684\u5f62\u5f0f\u8fd4\u56de\u7ed9 Driver,\u901a\u8fc7\u5c06\u8ba1\u7b97\u540e\u7684\u8f83\u5c0f\u7ed3\u679c\u96c6\u8fd4\u56de<\/td>\n<\/tr>\n
count()<\/td>\n\u8ba1\u7b97 RDD \u4e2d\u7684\u5143\u7d20\u4e2a\u6570<\/td>\n<\/tr>\n
first()<\/td>\n\u8fd4\u56de RDD \u4e2d\u7b2c\u4e00\u4e2a\u5143\u7d20<\/td>\n<\/tr>\n
take(n)<\/td>\n\u4ee5\u6570\u7ec4\u7684\u5f62\u5f0f\u8fd4\u56de RDD \u524d n \u4e2a\u5143\u7d20<\/td>\n<\/tr>\n
takeSample(withReplacement,num, [seed])<\/td>\n\u8fd4\u56de\u4e00\u4e2a\u6570\u7ec4\uff0c\u8be5\u6570\u7ec4\u7531\u4ece\u6570\u636e\u96c6\u4e2d\u968f\u673a\u91c7\u6837\u7684num\u4e2a\u5143\u7d20\u7ec4\u6210\uff0c\u53ef\u4ee5\u9009\u62e9\u662f\u5426\u7528\u968f\u673a\u6570\u66ff\u6362\u4e0d\u8db3\u7684\u90e8\u5206\uff0cseed\u7528\u4e8e\u6307\u5b9a\u968f\u673a\u6570\u751f\u6210\u5668\u79cd\u5b50<\/td>\n<\/tr>\n
saveAsTextFile(path)<\/td>\n\u5c06 RDD \u5b58\u50a8\u5230\u6587\u672c\u6587\u4ef6\u4e2d\uff0c\u5e76\u4e00\u6b21\u8c03\u7528\u6bcf\u4e2a\u5143\u7d20\u7684toString\u65b9\u6cd5\u5c06\u4e4b\u8f6c\u6362\u6210\u5b57\u7b26\u4e32\u4fdd\u5b58\u6210\u4e00\u884c<\/td>\n<\/tr>\n
saveAsSequenceFile(path)<\/td>\n\u9488\u5bf9 key\/value \u7c7b\u578b\u7684 RDD,\u4fdd\u5b58\u6210 SequenceFile \u683c\u5f0f\u6587\u4ef6<\/td>\n<\/tr>\n
countByKey()<\/td>\n\u9488\u5bf9 key\/value \u7c7b\u578b\u7684 RDD,\u7edf\u8ba1\u6bcf\u4e2a key\u51fa\u73b0\u7684\u6b21\u6570\uff0c\u5e76\u4ee5 hashmap \u5f62\u5f0f\u8fd4\u56de<\/td>\n<\/tr>\n
foreach(func)<\/td>\n\u5c06 RDD \u4e2d\u7684\u5143\u7d20\u4e00\u6b21\u4ea4\u7ed9 func \u5904\u7406<\/td>\n<\/tr>\n
aggregate<\/td>\n\u5148\u5bf9\u5206\u533a\u8fdb\u884c\u64cd\u4f5c\uff0c\u518d\u603b\u4f53\u64cd\u4f5c<\/td>\n<\/tr>\n
aggregateByKey<\/td>\n<\/td>\n<\/tr>\n
lookup(key: K)<\/td>\n\u9488\u5bf9 key\/value \u7c7b\u578b\u7684 RDD, \u6307\u5b9akey\u503c\uff0c\u8fd4\u56deRDD\u4e2d\u8be5K\u5bf9\u5e94\u7684\u6240\u6709V\u503c\u3002<\/td>\n<\/tr>\n
foreachPartition<\/td>\n\u7c7b\u4f3c\u4e8e foreach\uff0c\u4f46\u72ec\u7acb\u5730\u5728 RDD \u7684\u6bcf\u4e00\u4e2a\u5206\u7247\u4e0a\u8fd0\u884c\uff0c\u5176\u4e2d\u53ef\u5d4c\u5165foreach\u7b97\u5b50<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

<\/a>RDD\u5e38\u7528\u7684\u7b97\u5b50\u64cd\u4f5c\u6f14\u793a<\/h3>\n

\u4e3a\u4e86\u65b9\u4fbf\u524d\u671f\u7684\u6d4b\u8bd5\u548c\u5b66\u4e60\uff0c\u53ef\u4ee5\u4f7f\u7528spark-shell\u8fdb\u884c\u6f14\u793a<\/p>\n

\n\n\n
\n
1<\/span>
<\/pre>\n<\/td>\n
\n
spark-shell --master local[2]<\/span>
<\/pre>\n<\/td>\n<\/tr>\n<\/table>\n<\/figure>\n
<\/a>map<\/h4>\n
\n\n\n
\n
1<\/span>
2<\/span>
3<\/span>
4<\/span>
<\/pre>\n<\/td>\n
\n
val<\/span> rdd1 = sc.parallelize(List<\/span>(5<\/span>, 6<\/span>, 4<\/span>, 7<\/span>, 3<\/span>, 8<\/span>, 2<\/span>, 9<\/span>, 1<\/span>, 10<\/span>))<\/span>
<\/span>
\/\/\u628ardd1\u4e2d\u6bcf\u4e00\u4e2a\u5143\u7d20\u4e58\u4ee510<\/span><\/span>
rdd1.map(_*10<\/span>).collect<\/span>
<\/pre>\n<\/td>\n<\/tr>\n<\/table>\n<\/figure>\n
<\/a>filter<\/h4>\n
\n\n\n
\n
1<\/span>
2<\/span>
3<\/span>
4<\/span>
<\/pre>\n<\/td>\n
\n
val<\/span> rdd1 = sc.parallelize(List<\/span>(5<\/span>, 6<\/span>, 4<\/span>, 7<\/span>, 3<\/span>, 8<\/span>, 2<\/span>, 9<\/span>, 1<\/span>, 10<\/span>))<\/span>
<\/span>
\/\/\u628ardd1\u4e2d\u5927\u4e8e5\u7684\u5143\u7d20\u8fdb\u884c\u8fc7\u6ee4<\/span><\/span>
rdd1.filter(x => x >5<\/span>).collect<\/span>
<\/pre>\n<\/td>\n<\/tr>\n<\/table>\n<\/figure>\n
<\/a>flatMap<\/h4>\n
\n\n\n
\n
1<\/span>
2<\/span>
3<\/span>
<\/pre>\n<\/td>\n
\n
val<\/span> rdd1 = sc.parallelize(Array<\/span>(\"a b c\"<\/span>, \"d e f\"<\/span>, \"h i j\"<\/span>))<\/span>
\/\/\u83b7\u53d6rdd1\u4e2d\u5143\u7d20\u7684\u6bcf\u4e00\u4e2a\u5b57\u6bcd<\/span><\/span>
rdd1.flatMap(_.split(\" \"<\/span>)).collect<\/span>
<\/pre>\n<\/td>\n<\/tr>\n<\/table>\n<\/figure>\n
<\/a>intersection\u3001union<\/h4>\n
\n\n\n
\n
1<\/span>
2<\/span>
3<\/span>
4<\/span>
5<\/span>
6<\/span>
7<\/span>
<\/pre>\n<\/td>\n
\n
val<\/span> rdd1 = sc.parallelize(List<\/span>(5<\/span>, 6<\/span>, 4<\/span>, 3<\/span>))<\/span>
val<\/span> rdd2 = sc.parallelize(List<\/span>(1<\/span>, 2<\/span>, 3<\/span>, 4<\/span>))<\/span>
\/\/\u6c42\u4ea4\u96c6<\/span><\/span>
rdd1.intersection(rdd2).collect<\/span>
<\/span>
\/\/\u6c42\u5e76\u96c6<\/span><\/span>
rdd1.union(rdd2).collect<\/span>
<\/pre>\n<\/td>\n<\/tr>\n<\/table>\n<\/figure>\n
<\/a>distinct<\/h4>\n
\n\n\n
\n
1<\/span>
2<\/span>
3<\/span>
<\/pre>\n<\/td>\n
\n
val<\/span> rdd1 = sc.parallelize(List<\/span>(1<\/span>,1<\/span>,2<\/span>,3<\/span>,3<\/span>,4<\/span>,5<\/span>,6<\/span>,7<\/span>))<\/span>
\/\/\u53bb\u91cd<\/span><\/span>
rdd1.distinct<\/span>
<\/pre>\n<\/td>\n<\/tr>\n<\/table>\n<\/figure>\n
<\/a>join\u3001groupByKey<\/h4>\n
\n\n\n
\n
1<\/span>
2<\/span>
3<\/span>
4<\/span>
5<\/span>
6<\/span>
7<\/span>
8<\/span>
<\/pre>\n<\/td>\n
\n
val<\/span> rdd1 = sc.parallelize(List<\/span>((\"tom\"<\/span>, 1<\/span>), (\"jerry\"<\/span>, 3<\/span>), (\"kitty\"<\/span>, 2<\/span>)))<\/span>
val<\/span> rdd2 = sc.parallelize(List<\/span>((\"jerry\"<\/span>, 2<\/span>), (\"tom\"<\/span>, 1<\/span>), (\"shuke\"<\/span>, 2<\/span>)))<\/span>
\/\/\u6c42join<\/span><\/span>
val<\/span> rdd3 = rdd1.join(rdd2)<\/span>
rdd3.collect<\/span>
\/\/\u6c42\u5e76\u96c6<\/span><\/span>
val<\/span> rdd4 = rdd1 union rdd2<\/span>
rdd4.groupByKey.collect<\/span>
<\/pre>\n<\/td>\n<\/tr>\n<\/table>\n<\/figure>\n
<\/a>cogroup<\/h4>\n
\n\n\n
\n
1<\/span>
2<\/span>
3<\/span>
4<\/span>
5<\/span>
<\/pre>\n<\/td>\n
\n
val<\/span> rdd1 = sc.parallelize(List<\/span>((\"tom\"<\/span>, 1<\/span>), (\"tom\"<\/span>, 2<\/span>), (\"jerry\"<\/span>, 3<\/span>), (\"kitty\"<\/span>, 2<\/span>)))<\/span>
val<\/span> rdd2 = sc.parallelize(List<\/span>((\"jerry\"<\/span>, 2<\/span>), (\"tom\"<\/span>, 1<\/span>), (\"jim\"<\/span>, 2<\/span>)))<\/span>
\/\/\u5206\u7ec4<\/span><\/span>
val<\/span> rdd3 = rdd1.cogroup(rdd2)<\/span>
rdd3.collect<\/span>
<\/pre>\n<\/td>\n<\/tr>\n<\/table>\n<\/figure>\n
<\/a>reduce<\/h4>\n
\n\n\n
\n
1<\/span>
2<\/span>
3<\/span>
4<\/span>
5<\/span>
6<\/span>
7<\/span>
8<\/span>
9<\/span>
10<\/span>
<\/pre>\n<\/td>\n
\n
val<\/span> rdd1 = sc.parallelize(List<\/span>(1<\/span>, 2<\/span>, 3<\/span>, 4<\/span>, 5<\/span>))<\/span>
<\/span>
\/\/reduce\u805a\u5408<\/span><\/span>
val<\/span> rdd2 = rdd1.reduce(_ + _)<\/span>
rdd2.collect<\/span>
<\/span>
val<\/span> rdd3 = sc.parallelize(List<\/span>(\"1\"<\/span>,\"2\"<\/span>,\"3\"<\/span>,\"4\"<\/span>,\"5\"<\/span>))<\/span>
rdd3.reduce(_+_)<\/span>
<\/span>
\u8fd9\u91cc\u53ef\u80fd\u4f1a\u51fa\u73b0\u591a\u4e2a\u4e0d\u540c\u7684\u7ed3\u679c\uff0c\u7531\u4e8e\u5143\u7d20\u5728\u4e0d\u540c\u7684\u5206\u533a\u4e2d\uff0c\u6bcf\u4e00\u4e2a\u5206\u533a\u90fd\u662f\u4e00\u4e2a\u72ec\u7acb\u7684task\u7ebf\u7a0b\u53bb\u8fd0\u884c\u3002\u8fd9\u4e9btask\u8fd0\u884c\u6709\u5148\u540e\u5173\u7cfb<\/span>
<\/pre>\n<\/td>\n<\/tr>\n<\/table>\n<\/figure>\n
<\/a>reduceByKey\u3001sortByKey<\/h4>\n
\n\n\n
\n
1<\/span>
2<\/span>
3<\/span>
4<\/span>
5<\/span>
6<\/span>
7<\/span>
8<\/span>
9<\/span>
10<\/span>
11<\/span>
<\/pre>\n<\/td>\n
\n
val<\/span> rdd1 = sc.parallelize(List<\/span>((\"tom\"<\/span>, 1<\/span>), (\"jerry\"<\/span>, 3<\/span>), (\"kitty\"<\/span>, 2<\/span>),  (\"shuke\"<\/span>, 1<\/span>)))<\/span>
val<\/span> rdd2 = sc.parallelize(List<\/span>((\"jerry\"<\/span>, 2<\/span>), (\"tom\"<\/span>, 3<\/span>), (\"shuke\"<\/span>, 2<\/span>), (\"kitty\"<\/span>, 5<\/span>)))<\/span>
val<\/span> rdd3 = rdd1.union(rdd2)<\/span>
<\/span>
\/\/\u6309key\u8fdb\u884c\u805a\u5408<\/span><\/span>
val<\/span> rdd4 = rdd3.reduceByKey(_ + _)<\/span>
rdd4.collect<\/span>
<\/span>
\/\/\u6309value\u7684\u964d\u5e8f\u6392\u5e8f<\/span><\/span>
val<\/span> rdd5 = rdd4.map(t => (t._2, t._1)).sortByKey(false<\/span>).map(t => (t._2, t._1))<\/span>
rdd5.collect<\/span>
<\/pre>\n<\/td>\n<\/tr>\n<\/table>\n<\/figure>\n
<\/a>repartition\u3001coalesce<\/h4>\n
\n\n\n
\n
1<\/span>
2<\/span>
3<\/span>
4<\/span>
5<\/span>
6<\/span>
7<\/span>
8<\/span>
9<\/span>
10<\/span>
11<\/span>
12<\/span>
13<\/span>
14<\/span>
15<\/span>
16<\/span>
17<\/span>
18<\/span>
19<\/span>
20<\/span>
21<\/span>
22<\/span>
23<\/span>
24<\/span>
<\/pre>\n<\/td>\n
\n
val<\/span> rdd1 = sc.parallelize(1<\/span> to 10<\/span>,3<\/span>)<\/span>
\/\/\u6253\u5370rdd1\u7684\u5206\u533a\u6570<\/span><\/span>
rdd1.partitions.size<\/span>
<\/span>
\/\/\u5229\u7528repartition\u6539\u53d8rdd1\u5206\u533a\u6570<\/span><\/span>
\/\/\u51cf\u5c11\u5206\u533a<\/span><\/span>
rdd1.repartition(2<\/span>).partitions.size<\/span>
<\/span>
\/\/\u589e\u52a0\u5206\u533a<\/span><\/span>
rdd1.repartition(4<\/span>).partitions.size<\/span>
<\/span>
\/\/\u5229\u7528coalesce\u6539\u53d8rdd1\u5206\u533a\u6570<\/span><\/span>
\/\/\u51cf\u5c11\u5206\u533a<\/span><\/span>
rdd1.coalesce(2<\/span>).partitions.size<\/span>
<\/span>
<\/span>
\/\/repartition:  \u91cd\u65b0\u5206\u533a\uff0c \u6709shuffle<\/span><\/span>
\/\/coalesce:     \u5408\u5e76\u5206\u533a \/ \u51cf\u5c11\u5206\u533a \t\u9ed8\u8ba4\u4e0dshuffle   <\/span><\/span>
\/\/\u9ed8\u8ba4 coalesce \u4e0d\u80fd\u6269\u5927\u5206\u533a\u6570\u91cf\u3002\u9664\u975e\u6dfb\u52a0true\u7684\u53c2\u6570\uff0c\u6216\u8005\u4f7f\u7528repartition\u3002<\/span><\/span>
<\/span>
\/\/\u9002\u7528\u573a\u666f\uff1a<\/span><\/span>
\/\/1\u3001\u5982\u679c\u8981shuffle\uff0c\u90fd\u7528 repartition<\/span><\/span>
\/\/2\u3001\u4e0d\u9700\u8981shuffle\uff0c\u4ec5\u4ec5\u662f\u505a\u5206\u533a\u7684\u5408\u5e76\uff0ccoalesce<\/span><\/span>
\/\/3\u3001repartition\u5e38\u7528\u4e8e\u6269\u5927\u5206\u533a\u3002<\/span><\/span>
<\/pre>\n<\/td>\n<\/tr>\n<\/table>\n<\/figure>\n
<\/a>map\u3001mapPartitions\u3001mapPartitionsWithIndex<\/h4>\n
\n\n\n
\n
1<\/span>
2<\/span>
3<\/span>
4<\/span>
5<\/span>
6<\/span>
7<\/span>
8<\/span>
9<\/span>
10<\/span>
11<\/span>
12<\/span>
13<\/span>
<\/pre>\n<\/td>\n
\n
val<\/span> rdd1=sc.parallelize(1<\/span> to 10<\/span>,5<\/span>)<\/span>
rdd1.map(x => x*10<\/span>)).collect<\/span>
rdd1.mapPartitions(iter => iter.map(x=>x*10<\/span>)).collect<\/span>
<\/span>
\/\/index\u8868\u793a\u5206\u533a\u53f7  \u53ef\u4ee5\u83b7\u53d6\u5f97\u5230\u6bcf\u4e00\u4e2a\u5143\u7d20\u5c5e\u4e8e\u54ea\u4e00\u4e2a\u5206\u533a<\/span><\/span>
rdd1.mapPartitionsWithIndex((index,iter)=>iter.map(x=>(index,x)))<\/span>
<\/span>
map\uff1a\u7528\u4e8e\u904d\u5386RDD<\/span>,\u5c06\u51fd\u6570f\u5e94\u7528\u4e8e\u6bcf\u4e00\u4e2a\u5143\u7d20\uff0c\u8fd4\u56de\u65b0\u7684RDD<\/span>(transformation\u7b97\u5b50)\u3002<\/span>
mapPartitions:\u7528\u4e8e\u904d\u5386\u64cd\u4f5cRDD<\/span>\u4e2d\u7684\u6bcf\u4e00\u4e2a\u5206\u533a\uff0c\u8fd4\u56de\u751f\u6210\u4e00\u4e2a\u65b0\u7684RDD<\/span>\uff08transformation\u7b97\u5b50\uff09\u3002<\/span>
<\/span>
\u603b\u7ed3\uff1a<\/span>
\u5982\u679c\u5728\u6620\u5c04\u7684\u8fc7\u7a0b\u4e2d\u9700\u8981\u9891\u7e41\u521b\u5efa\u989d\u5916\u7684\u5bf9\u8c61\uff0c\u4f7f\u7528mapPartitions\u8981\u6bd4map\u9ad8\u6548<\/span>
\u6bd4\u5982\uff0c\u5c06RDD<\/span>\u4e2d\u7684\u6240\u6709\u6570\u636e\u901a\u8fc7JDBC<\/span>\u8fde\u63a5\u5199\u5165\u6570\u636e\u5e93\uff0c\u5982\u679c\u4f7f\u7528map\u51fd\u6570\uff0c\u53ef\u80fd\u8981\u4e3a\u6bcf\u4e00\u4e2a\u5143\u7d20\u90fd\u521b\u5efa\u4e00\u4e2aconnection\uff0c\u8fd9\u6837\u5f00\u9500\u5f88\u5927\uff0c\u5982\u679c\u4f7f\u7528mapPartitions\uff0c\u90a3\u4e48\u53ea\u9700\u8981\u9488\u5bf9\u6bcf\u4e00\u4e2a\u5206\u533a\u5efa\u7acb\u4e00\u4e2aconnection\u3002<\/span>
<\/pre>\n<\/td>\n<\/tr>\n<\/table>\n<\/figure>\n
<\/a>foreach\u3001foreachPartition<\/h4>\n
\n\n\n
\n
1<\/span>
2<\/span>
3<\/span>
4<\/span>
5<\/span>
6<\/span>
7<\/span>
8<\/span>
9<\/span>
10<\/span>
11<\/span>
12<\/span>
13<\/span>
14<\/span>
<\/pre>\n<\/td>\n
\n
val<\/span> rdd1 = sc.parallelize(List<\/span>(5<\/span>, 6<\/span>, 4<\/span>, 7<\/span>, 3<\/span>, 8<\/span>, 2<\/span>, 9<\/span>, 1<\/span>, 10<\/span>))<\/span>
<\/span>
\/\/foreach\u5b9e\u73b0\u5bf9rdd1\u91cc\u7684\u6bcf\u4e00\u4e2a\u5143\u7d20\u4e5810\u7136\u540e\u6253\u5370\u8f93\u51fa<\/span><\/span>
rdd1.foreach(x=>println(x * 10<\/span>))<\/span>
<\/span>
\/\/foreachPartition\u5b9e\u73b0\u5bf9rdd1\u91cc\u7684\u6bcf\u4e00\u4e2a\u5143\u7d20\u4e5810\u7136\u540e\u6253\u5370\u8f93\u51fa<\/span><\/span>
rdd1.foreachPartition(iter => iter.foreach(x=>println(x * 10<\/span>)))<\/span>
<\/span>
foreach:\u7528\u4e8e\u904d\u5386RDD<\/span>,\u5c06\u51fd\u6570f\u5e94\u7528\u4e8e\u6bcf\u4e00\u4e2a\u5143\u7d20\uff0c\u65e0\u8fd4\u56de\u503c(action\u7b97\u5b50)\u3002<\/span>
foreachPartition: \u7528\u4e8e\u904d\u5386\u64cd\u4f5cRDD<\/span>\u4e2d\u7684\u6bcf\u4e00\u4e2a\u5206\u533a\u3002\u65e0\u8fd4\u56de\u503c(action\u7b97\u5b50)\u3002<\/span>
<\/span>
<\/span>
\u603b\u7ed3\uff1a<\/span>
\u4e00\u822c\u4f7f\u7528mapPartitions\u6216\u8005foreachPartition\u7b97\u5b50\u6bd4map\u548cforeach\u66f4\u52a0\u9ad8\u6548\uff0c\u63a8\u8350\u4f7f\u7528\u3002<\/span>
<\/pre>\n<\/td>\n<\/tr>\n<\/table>\n<\/figure>\n","protected":false},"excerpt":{"rendered":"
RDD\u7684\u7b97\u5b50\u5206\u7c7b transfor…<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[],"tags":[],"_links":{"self":[{"href":"http:\/\/4563.org\/index.php?rest_route=\/wp\/v2\/posts\/142681"}],"collection":[{"href":"http:\/\/4563.org\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/4563.org\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/4563.org\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/4563.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=142681"}],"version-history":[{"count":0,"href":"http:\/\/4563.org\/index.php?rest_route=\/wp\/v2\/posts\/142681\/revisions"}],"wp:attachment":[{"href":"http:\/\/4563.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=142681"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/4563.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=142681"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/4563.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=142681"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}