{"id":7200,"date":"2025-09-30T15:37:37","date_gmt":"2025-09-30T06:37:37","guid":{"rendered":"https:\/\/blog.since2020.jp\/?p=7200"},"modified":"2025-09-30T15:37:37","modified_gmt":"2025-09-30T06:37:37","slug":"sparksql_glossary","status":"publish","type":"post","link":"https:\/\/since2020.jp\/media\/sparksql_glossary\/","title":{"rendered":"SparkSQL\u3092\u300c\u3044\u307e\u300d\u4f7f\u3044\u3053\u306a\u3059\u20142025\u5e74\u7248\u30fb\u5b9f\u8df5\u304b\u3089\u5b66\u3076\u8a2d\u8a08\u601d\u60f3\u3068\u6700\u65b0\u30c8\u30ec\u30f3\u30c9"},"content":{"rendered":"\n<p>SparkSQL\u306e\u57fa\u672c\u6982\u5ff5\u304b\u3089\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3001\u30d7\u30ed\u30bb\u30b9\u3001\u5b9f\u8df5\u624b\u6cd5\u3001\u6700\u65b0\u30c8\u30ec\u30f3\u30c9\uff08Spark 4.0\u30fbANSI\u30e2\u30fc\u30c9\u65e2\u5b9a\u5316\u30fbSpark Connect\uff09\u307e\u3067\u3092\u3001\u30c7\u30fc\u30bf\u57fa\u76e4\u306e\u610f\u601d\u6c7a\u5b9a\u306b\u5f79\u7acb\u3064\u89b3\u70b9\u3067\u89e3\u8aac\u3057\u307e\u3059\u3002\u30e6\u30fc\u30b9\u30b1\u30fc\u30b9\u3084\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u306e\u52d8\u6240\u3082\u7db2\u7f85\u3057\u3001\u30d3\u30b8\u30cd\u30b9\u3068\u7814\u7a76\u306e\u53cc\u65b9\u306b\u52b9\u304f\u77e5\u898b\u3092\u63d0\u4f9b\u3002<\/p>\n\n\n<h2>\u306a\u305c\u30012025\u5e74\u306e\u4eca\u3042\u3089\u305f\u3081\u3066\u300cSparkSQL\u300d\u304b<\/h2>\n<p>\u30af\u30e9\u30a6\u30c9\u306b\u76f4\u7d50\u3057\u305f\u30ec\u30a4\u30af\u30cf\u30a6\u30b9\u306e\u63a1\u7528\u3067\u300c<strong>SQL\u3067\u5de8\u5927\u30c7\u30fc\u30bf\u3092\u901f\u304f\u30fb\u5b89\u5168\u306b\u6271\u3046<\/strong>\u300d\u3053\u3068\u304c\u5f53\u305f\u308a\u524d\u306b\u306a\u308a\u307e\u3057\u305f\u3002SparkSQL\u306f\u3001\u305d\u306e\u524d\u63d0\u3092\u652f\u3048\u308b<strong>\u5ba3\u8a00\u7684API<\/strong>\uff08SQL\/DataFrame\/Dataset\uff09\u3068<strong>\u5b9f\u884c\u6642\u6700\u9069\u5316<\/strong>\uff08Catalyst\/AQE\/Tungsten\uff09\u3092\u517c\u306d\u5099\u3048\u3001<strong>ETL\u30fbBI\u30fbML \u524d\u51e6\u7406\u30fb\u30b9\u30c8\u30ea\u30fc\u30df\u30f3\u30b0<\/strong>\u3092\u4e00\u3064\u306e\u62bd\u8c61\u5316\u306b\u307e\u3068\u3081\u3042\u3052\u307e\u3059\u3002Databricks \u5b9f\u88c5\u306e\u73fe\u5834\u8a18\u4e8b\u3067\u3082\u3001<strong>\u30e1\u30bf\u30b9\u30c8\u30a2\uff08Hive\/Unity Catalog\uff09<\/strong>\u304c\u30c6\u30fc\u30d6\u30eb\u7ba1\u7406\u306e\u8981\u3067\u3042\u308b\u3053\u3068\u3001SQL\u306e\u307f\u3067\u30c6\u30fc\u30d6\u30eb\u4f5c\u6210\u304b\u3089\u30af\u30a8\u30ea\u3001\u6a29\u9650\u7ba1\u7406\u307e\u3067\u5b8c\u7d50\u3067\u304d\u308b\u4f53\u9a13\u304c\u793a\u3055\u308c\u3066\u3044\u307e\u3059\u3002<\/p>\r\n<p>\u3053\u306e\u300cSQL\u4e2d\u5fc3\u30fb\u30ac\u30d0\u30ca\u30f3\u30b9\u4e00\u4f53\u300d\u306e\u6f6e\u6d41\u306f\u3001<strong>Spark 4.0<\/strong>\u3067\u3055\u3089\u306b\u52a0\u901f\u3002<strong>SQL\u30b9\u30af\u30ea\u30d7\u30c8\u306e\u5236\u5fa1\u69cb\u6587\u3084\u518d\u5229\u7528\u53ef\u80fd\u306aUDF\u3001PySpark\u306eUDTF<\/strong>\u306a\u3069\u3001\u8a00\u8a9e\u6a5f\u80fd\u3068\u958b\u767a\u4f53\u9a13\u306e\u4e21\u8f2a\u304c\u62e1\u5f35\u3055\u308c\u307e\u3057\u305f\u3002\u3055\u3089\u306b<strong>Spark Connect<\/strong>\u306f\u30af\u30e9\u30a4\u30a2\u30f3\u30c8\u30fb\u30b5\u30fc\u30d0\u5206\u96e2\u3067\u958b\u767a\u8005\u4f53\u9a13\u3068\u62e1\u5f35\u6027\u3092\u5e95\u4e0a\u3052\u3057\u3066\u3044\u307e\u3059\u3002<\/p>\r\n<p><!-- notionvc: ef0151ab-05b6-4249-84e9-9e3999cedc3d --><\/p>\n\n<h2>SparkSQL\u306e\u30b3\u30a2\u6982\u5ff5\u3092\u3001\u8a2d\u8a08\u601d\u60f3\u304b\u3089\u7406\u89e3\u3059\u308b<\/h2>\n<b>\u300c\u5ba3\u8a00\u300d\u3092\u6700\u9069\u5316\u306b\u843d\u3068\u3057\u8fbc\u3080\uff1aCatalyst\u30fbTungsten\u30fbAQE<\/b>\r\n<p>SparkSQL\u306f\u3001SQL\u3084DataFrame\u3067\u66f8\u304b\u308c\u305f<strong>\u5ba3\u8a00\u7684\u306a\u76ee\u7684<\/strong>\u3092\u3001<strong>Analyzer \u2192 Optimizer \u2192 Planner \u2192 \u7269\u7406\u8a08\u753b \u2192 JVM\u30b3\u30fc\u30c9\u751f\u6210<\/strong>\u3068\u3044\u3046\u30b3\u30f3\u30d1\u30a4\u30eb\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u3067\u5b9f\u884c\u3078\u843d\u3068\u3057\u8fbc\u307f\u307e\u3059\u3002Databricks\u30b3\u30df\u30c3\u30bf\u30fc\u306e\u89e3\u8aac\u3067\u306f\u3001\u30af\u30a8\u30ea\u304c<strong>\u672a\u89e3\u6c7a\u306e\u8ad6\u7406\u8a08\u753b<\/strong>\u304b\u3089\u30e1\u30bf\u30c7\u30fc\u30bf\u3092\u901a\u3058\u3066\u610f\u5473\u4ed8\u3051\u3055\u308c\u3001\u7d4c\u9a13\u5247\u3084\u30b3\u30b9\u30c8\u306b\u57fa\u3065\u304f\u6700\u9069\u5316\u3092\u53d7\u3051\u3066RDD\u5b9f\u884c\u3078\u5909\u63db\u3055\u308c\u308b\u6d41\u308c\u304c\u8a73\u8ff0\u3055\u308c\u3066\u3044\u307e\u3059\u3002<strong>EXPLAIN<\/strong>\u3067\u30d7\u30e9\u30f3\u3092\u8aad\u3080\u3053\u3068\u304c\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u306e\u51fa\u767a\u70b9\u3067\u3042\u308b\u3053\u3068\u3082\u5f37\u8abf\u3055\u308c\u3066\u3044\u307e\u3059\u3002<\/p>\r\n<p>2025\u5e74\u6642\u70b9\u3067\u3082\u3001<strong>Catalyst\u6700\u9069\u5316<\/strong>\uff08\u8ff0\u8a9e\u4e0b\u63a8\u3057\u30fb\u6295\u5f71\u524a\u6e1b\u30fb\u7d50\u5408\u9806\u5e8f\u6700\u9069\u5316\uff09\u3068<strong>Tungsten<\/strong>\uff08\u30e1\u30e2\u30ea\u52b9\u7387\u30fb\u30b3\u30fc\u30c9\u751f\u6210\uff09\u306e\u8a2d\u8a08\u306f\u4e0d\u5909\u306e\u5f37\u307f\u3002\u3055\u3089\u306b<strong>AQE\uff08Adaptive Query Execution\uff09<\/strong>\u304c\u5b9f\u884c\u6642\u7d71\u8a08\u3067\u7d50\u5408\u6226\u7565\u3084\u30d1\u30fc\u30c6\u30a3\u30b7\u30e7\u30f3\u6570\u3092\u52d5\u7684\u306b\u8abf\u6574\u3057\u3001\u30b9\u30ad\u30e5\u30fc\u3068\u30b7\u30e3\u30c3\u30d5\u30eb\u306e\u75db\u70b9\u3092\u7de9\u548c\u3057\u307e\u3059\u3002Spark 3.5\u7cfb\u3067\u306fAQE\u306e\u9069\u7528\u7bc4\u56f2\u3082\u5e83\u304c\u308a\u307e\u3057\u305f\u3002<\/p>\r\n<b>\u30e1\u30bf\u30b9\u30c8\u30a2\u3068\u30ab\u30bf\u30ed\u30b0\uff1aHive\u304b\u3089Unity Catalog\u3078<\/b>\r\n<p>Qiita\u306e\u8a18\u4e8b\u304c\u793a\u3059\u901a\u308a\u3001SparkSQL\u306e\u300c\u30c6\u30fc\u30d6\u30eb\u540d\u300d\u3092\u610f\u5473\u3065\u3051\u308b\u306e\u306f<strong>\u30e1\u30bf\u30b9\u30c8\u30a2<\/strong>\u3067\u3059\u3002Databricks \u3067\u306f<strong>Unity Catalog<\/strong>\u304c\u7d71\u5408\u30ab\u30bf\u30ed\u30b0\u3068\u3057\u3066\u3001\u540d\u524d\u89e3\u6c7a\u30fb\u6a29\u9650\u30fb\u76e3\u67fb\u30fb\u30ea\u30cd\u30fc\u30b8\u30fb\u54c1\u8cea\u76e3\u8996\u3092\u62c5\u3044\u307e\u3059\u30022025\u5e74\u6642\u70b9\u306e\u516c\u5f0f\u30c9\u30ad\u30e5\u30e1\u30f3\u30c8\u3067\u3082\u3001<strong>\u65e7\u6765\u306eHive\u30e1\u30bf\u30b9\u30c8\u30a2\u4e2d\u5fc3\u306e\u30a2\u30af\u30bb\u30b9\u5236\u5fa1\u306f\u30ec\u30ac\u30b7\u30fc\u6271\u3044<\/strong>\u3068\u306a\u308a\u3001Unity Catalog\u3078\u306e\u79fb\u884c\u304c\u63a8\u5968\u3055\u308c\u3066\u3044\u307e\u3059\u3002<\/p>\r\n<p><!-- notionvc: 3c819062-74c2-40e3-8786-6e0e4bd2c175 --><\/p>\n\n<h2>\u300c\u6700\u65b0\u306eSparkSQL\u300d\u3092\u6b63\u3057\u304f\u4f7f\u3046\u305f\u3081\u306e\u30a2\u30c3\u30d7\u30c7\u30fc\u30c8\u8981\u70b9\uff082025\uff09<\/h2>\n<b>ANSI\u30e2\u30fc\u30c9\u306e\u65e2\u5b9a\u5316\u3067\u300c\u52d5\u304f\u3051\u3069\u5371\u306a\u3044\u300d\u3092\u9632\u3050<\/b>\r\n<p><strong>Spark 4.0 \u304b\u3089 ANSI \u6e96\u62e0\u304c\u65e2\u5b9a<\/strong>\u306b\u306a\u308a\u307e\u3057\u305f\u3002\u3053\u308c\u306b\u3088\u308a\u3001\u6697\u9ed9\u306e\u578b\u5909\u63db\u3084\u30b5\u30a4\u30ec\u30f3\u30c8\u306a\u30c7\u30fc\u30bf\u5207\u308a\u6368\u3066\u306b\u8d77\u56e0\u3059\u308b\u4e8b\u6545\u3092\u9632\u304e\u3001RDB\u3068\u306e\u79fb\u884c\u4e92\u63db\u3082\u53d6\u308a\u3084\u3059\u304f\u306a\u308a\u307e\u3059\u3002\u5fc5\u8981\u306a\u3089\u30ec\u30ac\u30b7\u30fc\u6319\u52d5\u3078\u623b\u3059\u8a2d\u5b9a\u3082\u63d0\u793a\u3055\u308c\u3066\u3044\u307e\u3059\u3002<strong>\u79fb\u884c\u6642\u306f\u30ad\u30e3\u30b9\u30c8\u3068\u95a2\u6570\u4e92\u63db\u6027\u306e\u56de\u5e30\u30c6\u30b9\u30c8<\/strong>\u3092\u7528\u610f\u3057\u307e\u3057\u3087\u3046\u3002<\/p>\r\n<b>Spark Connect \u3068\u30af\u30e9\u30a4\u30a2\u30f3\u30c8\u958b\u767a\u4f53\u9a13\u306e\u5237\u65b0<\/b>\r\n<p><strong>Spark Connect<\/strong>\u306f\u30af\u30e9\u30a4\u30a2\u30f3\u30c8\u3068\u30b5\u30fc\u30d0\u3092gRPC\u8d8a\u3057\u306b\u5206\u96e2\u3057\u3001<strong>\u8ad6\u7406\u8a08\u753b\u3092Arrow\u30d0\u30c3\u30c1\u3067\u30b9\u30c8\u30ea\u30fc\u30e0\u8fd4\u5374<\/strong>\u3059\u308b\u8a2d\u8a08\u3002\u8efd\u91cf\u30af\u30e9\u30a4\u30a2\u30f3\u30c8\u3067\u306e\u958b\u767a\u3084\u591a\u8a00\u8a9e\u5bfe\u5fdc\u3092\u63a8\u9032\u3057\u30014.0\u3067\u306fAPI\u306e\u62e1\u5145\u3084\u30e2\u30fc\u30c9\u5207\u66ff\u3001\u8907\u6570\u8a00\u8a9e\u30af\u30e9\u30a4\u30a2\u30f3\u30c8\u306e\u6574\u5099\u304c\u9032\u307f\u307e\u3057\u305f\u3002\u30ce\u30fc\u30c8\u30d6\u30c3\u30af\u4ee5\u5916\u306e\u30a2\u30d7\u30ea\u3084\u30b5\u30fc\u30d3\u30b9\u304b\u3089\u3082\u3001\u3088\u308a\u5b89\u5168\u306b\u30b9\u30b1\u30fc\u30eb\u3059\u308b\u300c<strong>\u758e\u7d50\u5408\u306aSpark<\/strong>\u300d\u3092\u5b9f\u73fe\u3067\u304d\u307e\u3059\u3002<\/p>\r\n<b>SQL\u8a00\u8a9e\u6a5f\u80fd\u30fbPython\u9023\u643a\u306e\u5f37\u5316<\/b>\r\n<p>Spark 4.0\u3067\u306f<strong>SQL\u30b9\u30af\u30ea\u30d7\u30c8\u306e\u5909\u6570\u30fb\u5236\u5fa1\u69cb\u6587\u3001\u518d\u5229\u7528UDF\u3001PIPE\u69cb\u6587<\/strong>\u306a\u3069\u304c\u8ffd\u52a0\u3055\u308c\u3001\u8907\u96d1\u306a\u5206\u6790\u30d5\u30ed\u30fc\u3092<strong>SQL\u3060\u3051\u3067\u7d44\u307f\u7acb\u3066\u3084\u3059\u304f<\/strong>\u306a\u308a\u307e\u3057\u305f\u3002PySpark\u3067\u306f<strong>Python UDTF<\/strong>\u3084<strong>UDF\u30d7\u30ed\u30d5\u30a1\u30a4\u30ea\u30f3\u30b0\u306e\u7d71\u5408<\/strong>\u3001\u8efd\u91cf\u30af\u30e9\u30a4\u30a2\u30f3\u30c8\u306a\u3069\u958b\u767a\u8005\u4f53\u9a13\u304c\u6539\u5584\u3002\u5927\u898f\u6a21\u7279\u5fb4\u91cf\u751f\u6210\u3084\u30e2\u30cb\u30bf\u30ea\u30f3\u30b0\u3067\u306e\u300c<strong>Python\u306e\u6a5f\u52d5\u529b\uff0b\u5206\u6563SQL<\/strong>\u300d\u306e\u7d44\u307f\u5408\u308f\u305b\u304c\u3088\u308a\u73fe\u5b9f\u7684\u306b\u3002<\/p>\r\n<p><!-- notionvc: 8b3c58fe-cfd4-45bc-9541-dbecde62defe --><\/p>\n\n<h2>\u30d7\u30ed\u30bb\u30b9\u3068\u624b\u6cd5\uff1a\u73fe\u5834\u3067\u306e\u300c\u52dd\u3061\u30d1\u30bf\u30fc\u30f3\u300d<\/h2>\n<b>\u6a19\u6e96\u30d7\u30ed\u30bb\u30b9\uff08\u6700\u77ed\u30eb\u30fc\u30c8\uff09<\/b>\r\n<ol>\r\n\t<li><strong>SparkSession \u6e96\u5099<\/strong>\uff08Connect\/\u30af\u30e9\u30b9\u30bf\u8a2d\u5b9a\uff0fANSI\u78ba\u8a8d\uff09<\/li>\r\n\t<li><strong>\u8aad\u307f\u8fbc\u307f<\/strong>\uff1a<code>spark.read<\/code>\u3067<strong>\u5217\u6307\u5411\uff08Parquet\/ORC\uff09<\/strong>\u3092\u512a\u5148\u3002\u30b9\u30ad\u30fc\u30de\u306f\u660e\u793a\u3002<\/li>\r\n\t<li><strong>\u30d3\u30e5\u30fc\/\u30c6\u30fc\u30d6\u30eb\u5316<\/strong>\uff1a<code>CREATE TABLE<\/code>\/<code>CREATE VIEW<\/code> \u304b <code>createOrReplaceTempView<\/code> \u3067<strong>\u518d\u5229\u7528\u5358\u4f4d<\/strong>\u3092\u4f5c\u308b\u3002<\/li>\r\n\t<li><strong>\u6700\u9069\u5316<\/strong>\uff1a<strong>\u30d1\u30fc\u30c6\u30a3\u30b7\u30e7\u30f3\u8a2d\u8a08\u30fb\u7d71\u8a08\u53ce\u96c6\u30fbAQE\u30aa\u30f3\u30fb\u30ad\u30e3\u30c3\u30b7\u30e5\u306f\u30d4\u30f3\u30dd\u30a4\u30f3\u30c8<\/strong>\u3002<\/li>\r\n\t<li><strong>\u51fa\u529b<\/strong>\uff1a<strong>\u7ba1\u7406\u30c6\u30fc\u30d6\u30eb\uff08Unity Catalog\uff09\u304b\u5916\u90e8\u30b7\u30f3\u30af\u3078\u3002\u30ac\u30d0\u30ca\u30f3\u30b9\u8981\u4ef6\u306fUC\u57fa\u6e96\u306b\u5408\u308f\u305b\u308b\u3002 Databricks\u3067\u306e\u4e00\u9023\u306e\u64cd\u4f5c\u306f\u3001\u30ce\u30fc\u30c8\u30d6\u30c3\u30af\u304b\u3089SQL\u5358\u4f53<\/strong>\u3067\u3082\u5b8c\u7d50\u3067\u304d\u307e\u3059\u3002<\/li>\r\n<\/ol>\r\n<b>DataFrame\/SQL\u306e\u4f7f\u3044\u5206\u3051<\/b>\r\n<p><strong>SQL<\/strong>\u306f\u96c6\u7d04\u3084\u7d50\u5408\u304c\u4e2d\u5fc3\u306e<strong>\u5206\u6790\u30ed\u30b8\u30c3\u30af<\/strong>\u3067\u3001<strong>DataFrame<\/strong>\u306f<strong>\u578b\u5b89\u5168\u3084\u95a2\u6570\u5408\u6210<\/strong>\u304c\u6b32\u3057\u3044\u3068\u304d\u306b\u6709\u5229\u3002\u6700\u9069\u5316\uff08Catalyst\/AQE\uff09\u306f<strong>\u3069\u3061\u3089\u3067\u3082\u52b9\u304f<\/strong>\u305f\u3081\u3001<strong>\u30c1\u30fc\u30e0\u306e\u53ef\u8aad\u6027<\/strong>\u3092\u57fa\u6e96\u306b\u9078\u3076\u306e\u304c\u5b9f\u52d9\u7684\u3067\u3059\u3002<strong>EXPLAIN<\/strong>\u3084<code>queryExecution<\/code>\u3067\u7269\u7406\u8a08\u753b\u3092\u5e38\u306b\u89b3\u5bdf\u3057\u307e\u3057\u3087\u3046\u3002<\/p>\r\n<b>\u300c\u672c\u5f53\u306b\u52b9\u304f\u300d\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u306e\u9806\u5e8f<\/b>\r\n<ul>\r\n\t<li><strong>I\/O\u306e\u52dd\u3061\u7b4b<\/strong>\uff1a<strong>\u5217\u6307\u5411\uff0b\u30ab\u30e9\u30e0\u526a\u5b9a\uff0b\u30d5\u30a3\u30eb\u30bf\u4e0b\u63a8\u3057<\/strong>\u3002\u307e\u305a<strong>\u30d5\u30a9\u30fc\u30de\u30c3\u30c8\u3068\u30d1\u30fc\u30c6\u30a3\u30b7\u30e7\u30f3\u8a2d\u8a08<\/strong>\u30678\u5272\u6c7a\u307e\u308b\u3002<\/li>\r\n\t<li><strong>\u7d71\u8a08\u3068\u7d50\u5408\u9806\u5e8f<\/strong>\uff1a\u8868\u7d71\u8a08\u304c\u7121\u3044\u3068Optimizer\u306f\u76f2\u76ee\u3002<strong>ANALYZE TABLE<\/strong>\u306e\u904b\u7528\u3092\u30eb\u30fc\u30c1\u30f3\u5316\u3002<\/li>\r\n\t<li><strong>AQE\u306e\u6d3b\u7528<\/strong>\uff1a\u30b9\u30ad\u30e5\u30fc\u89e3\u6d88\u30fb\u52d5\u7684\u7d50\u5408\u5207\u66ff\u30fb\u52d5\u7684\u30d1\u30fc\u30c6\u30a3\u30b7\u30e7\u30f3\u8abf\u6574\u3092<strong>\u65e2\u5b9a\u30aa\u30f3<\/strong>\u3067\u4f7f\u3044\u3053\u306a\u3059\u3002<\/li>\r\n\t<li><strong>\u30ad\u30e3\u30c3\u30b7\u30e5\u306f\u4e07\u80fd\u3067\u306f\u306a\u3044<\/strong>\uff1a\u3080\u3057\u308d\u9045\u304f\u306a\u308b\u5834\u5408\u3042\u308a\u3002<strong>\u5fc5\u8981\u7b87\u6240\u3060\u3051\u30d4\u30f3\u30dd\u30a4\u30f3\u30c8<\/strong>\u306b\u3002<\/li>\r\n\t<li><strong>ANSI\u65e2\u5b9a\u5316\u306e\u5f71\u97ff<\/strong>\uff1a\u6697\u9ed9\u30ad\u30e3\u30b9\u30c8\u524d\u63d0\u306e\u30b8\u30e7\u30d6\u306f\u843d\u3061\u308b\u3002<strong>\u578b\u3068NULL\u51e6\u7406<\/strong>\u3092\u5148\u306b\u662f\u6b63\u3057\u3001<strong>\u79fb\u884c\u30ac\u30fc\u30c9<\/strong>\u3092\u5165\u308c\u308b\u3002<\/li>\r\n<\/ul>\r\n<p><!-- notionvc: 4bf2c0ff-6d82-4ed6-a5b9-98a32a9c6850 --><\/p>\n\n<h2>\u30e6\u30fc\u30b9\u30b1\u30fc\u30b9\u3067\u5b66\u3076\uff1a\u3069\u3053\u306b\u52b9\u304f\u306e\u304b<\/h2>\n<b>\u30ec\u30a4\u30af\u30cf\u30a6\u30b9ETL\u3068\u30c7\u30fc\u30bf\u54c1\u8cea<\/b>\r\n<p>\u30aa\u30d6\u30b8\u30a7\u30af\u30c8\u30b9\u30c8\u30ec\u30fc\u30b8\u4e0a\u306e<strong>Parquet\uff0fORC<\/strong>\u3084<strong>\u30c6\u30fc\u30d6\u30eb\u5f62\u5f0f\uff08Delta\/Iceberg\/Hudi\uff09\u3092\u3001SparkSQL\u3067\u4e00\u8cab\u51e6\u7406\u3002\u30bf\u30a4\u30e0\u30c8\u30e9\u30d9\u30eb\u30fbACID\u30fb\u30b9\u30ad\u30fc\u30de\u9032\u5316\u306a\u3069\u30c6\u30fc\u30d6\u30eb\u5f62\u5f0f\u306e\u5229\u70b9\u3068\u3001Unity Catalog\u306e\u6a29\u9650\u30fb\u76e3\u67fb\u30fb\u30ea\u30cd\u30fc\u30b8<\/strong>\u3092\u7d44\u307f\u5408\u308f\u305b\u308b\u3068\u3001<strong>\u76e3\u67fb\u53ef\u80fd\u306a\u30c7\u30fc\u30bf\u57fa\u76e4<\/strong>\u3092SQL\u4e2d\u5fc3\u3067\u904b\u7528\u3067\u304d\u307e\u3059\u3002<\/p>\r\n<b>BI\u30fb\u30a2\u30c9\u30db\u30c3\u30af\u5206\u6790\u3068\u30c7\u30fc\u30bf\u30ac\u30d0\u30ca\u30f3\u30b9<\/b>\r\n<p>\u30c0\u30c3\u30b7\u30e5\u30dc\u30fc\u30c9\u306e\u88cf\u5074\u3067SparkSQL\u3092\u4f7f\u3046\u306a\u3089\u3001<strong>\u7ba1\u7406\u30c6\u30fc\u30d6\u30eb\uff08Unity Catalog Managed Tables\uff09<\/strong>\u304c\u65e2\u5b9a\u3002\u30af\u30e9\u30a6\u30c9\u30fb\u30e1\u30bf\u30c7\u30fc\u30bf\u30fb\u6700\u9069\u5316\u304c\u7d71\u5408\u3055\u308c\u3001<strong>\u30b3\u30b9\u30c8\u3068\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u3092\u5b66\u7fd2\u7684\u306b\u6700\u9069\u5316<\/strong>\u3059\u308b\u8a18\u8ff0\u3082\u51fa\u3066\u304d\u3066\u3044\u307e\u3059\u3002<strong>\u30ec\u30a4\u30af\u30cf\u30a6\u30b9\u00d7\u30ac\u30d0\u30ca\u30f3\u30b9<\/strong>\u306e\u5b9f\u88c5\u306f\u300cBI\u304c\u5b89\u5fc3\u3057\u3066\u4f7f\u3048\u308bSQL\u300d\u3092\u73fe\u5b9f\u306b\u3057\u307e\u3059\u3002<\/p>\r\n<b>ML\u524d\u51e6\u7406\u30fb\u7279\u5fb4\u91cf\u57fa\u76e4\u30fb\u30b9\u30c8\u30ea\u30fc\u30df\u30f3\u30b0<\/b>\r\n<p>\u7279\u5fb4\u91cf\u751f\u6210\u3084\u96c6\u8a08\u306e\u591a\u304f\u306f<strong>SQL\u306e\u30a6\u30a3\u30f3\u30c9\u30a6\u95a2\u6570\u3084\u96c6\u7d04<\/strong>\u3067\u8ff0\u8a9e\u5316\u3067\u304d\u3001<strong>PySpark UDTF\/UDF<\/strong>\u3067\u62e1\u5f35\u6027\u3082\u78ba\u4fdd\u3002<strong>Structured Streaming<\/strong>\u306b\u3088\u308a\u3001\u30d0\u30c3\u30c1\u3068\u540c\u3058\u62bd\u8c61\u3067<strong>\u8fd1\u30ea\u30a2\u30eb\u30bf\u30a4\u30e0<\/strong>\u306b\u7279\u5fb4\u91cf\u3092\u66f4\u65b0\u3067\u304d\u307e\u3059\u3002Spark 4.0\u3067\u306f<strong>\u72b6\u614b\u7ba1\u7406\u3084\u30c7\u30d0\u30c3\u30b0\u6027<\/strong>\u3082\u5f37\u5316\u3055\u308c\u3066\u3044\u307e\u3059\u3002<\/p>\r\n<p><!-- notionvc: ded84b68-deb7-4c46-b613-76275e1196a7 --><\/p>\n\n<h2>\u843d\u3068\u3057\u7a74\u3068\u30a2\u30f3\u30c1\u30d1\u30bf\u30fc\u30f3\uff08\u5bfe\u7b56\u3064\u304d\uff09<\/h2>\n<b>\u5c0f\u30d5\u30a1\u30a4\u30eb\u5730\u7344\u3068\u30b9\u30ad\u30e5\u30fc<\/b>\r\n<p>\u5c0f\u30d5\u30a1\u30a4\u30eb\u4e71\u7acb\u306f\u30b7\u30e3\u30c3\u30d5\u30eb\u5897\u3068\u30e1\u30bf\u30c7\u30fc\u30bf\u8ca0\u8377\u3092\u62db\u304d\u307e\u3059\u3002<strong>\u30aa\u30fc\u30c8\u30aa\u30d7\u30c6\u30a3\u30de\u30a4\u30ba\uff0f\u6700\u9069\u51fa\u529b\u30b5\u30a4\u30ba<\/strong>\u3001\u96c6\u8a08\u524d\u306e<strong>coalesce\/repartition<\/strong>\u3001<strong>ZORDER\/\u30af\u30e9\u30b9\u30bf\u30ea\u30f3\u30b0<\/strong>\uff08\u5b9f\u88c5\u4f9d\u5b58\uff09\u3067\u56de\u907f\u3002\u7d50\u5408\u30b9\u30ad\u30e5\u30fc\u306f<strong>AQE\u306e\u30b9\u30ad\u30e5\u30fc\u5206\u5272<\/strong>\u3068<strong>\u30d6\u30ed\u30fc\u30c9\u30ad\u30e3\u30b9\u30c8\u7d50\u5408\u306e\u95be\u5024\u8abf\u6574<\/strong>\u3067\u523a\u3059\u3002<\/p>\r\n<b>\u30ad\u30e3\u30c3\u30b7\u30e5\u4e07\u80fd\u8aac<\/b>\r\n<p>\u30ad\u30e3\u30c3\u30b7\u30e5\u306f<strong>\u9045\u304f\u306a\u308b<\/strong>\u3053\u3068\u304c\u3042\u308a\u307e\u3059\u3002\u30e1\u30e2\u30ea\u5727\u8feb\u2192\u30c7\u30a3\u30b9\u30af\u9000\u907f\u2192\u518d\u8aad\u8fbc\u306e\u30aa\u30fc\u30d0\u30fc\u30d8\u30c3\u30c9\u306b\u6ce8\u610f\u3002<strong>\u5fc5\u8981\u7b87\u6240\u3060\u3051\u3001\u77ed\u547d\u306e\u518d\u5229\u7528<\/strong>\u306b\u9650\u308b\u306e\u304c\u30b3\u30c4\u3002<\/p>\r\n<b>\u30b9\u30c8\u30a2\u3068\u6a29\u9650\u306e\u5206\u6563<\/b>\r\n<p><strong>Hive\u30e1\u30bf\u30b9\u30c8\u30a2\uff0bACL<\/strong>\u306e\u5bc4\u305b\u96c6\u3081\u904b\u7528\u306f\u30012025\u5e74\u306e\u898f\u6a21\u3067\u306f<strong>\u30ea\u30b9\u30af<\/strong>\u3002<strong>Unity Catalog\u3078\u6a19\u6e96\u5316<\/strong>\u3057\u3001<strong>\u30ab\u30bf\u30ed\u30b0\uff0f\u30b9\u30ad\u30fc\u30de\uff0f\u30c6\u30fc\u30d6\u30eb<\/strong>\u306e3\u968e\u5c64\u547d\u540d\u3068<strong>\u30c7\u30fc\u30bf\u6240\u6709\u6a29<\/strong>\u3092\u660e\u78ba\u5316\u3002<strong>\u30c7\u30fc\u30bf\u767a\u898b\u30fb\u30ea\u30cd\u30fc\u30b8\u30fb\u76e3\u67fb<\/strong>\u307e\u3067\u4e00\u8cab\u3067\u3002<\/p>\r\n<b>RDB\u79fb\u884c\u6642\u306e\u300c\u9ed9\u3063\u3066\u901a\u308b\u300d\u5730\u96f7<\/b>\r\n<p>4.0\u3067ANSI\u65e2\u5b9a\u5316\u3002<strong>\u6570\u5024\u30aa\u30fc\u30d0\u30fc\u30d5\u30ed\u30fc\u30fb\u6697\u9ed9\u30ad\u30e3\u30b9\u30c8<\/strong>\u30fb\u65e5\u4ed8\u578b\u306e\u4e92\u63db\u304c\u300c\u9ed9\u3063\u3066\u901a\u3089\u306a\u3044\u300d\u4e16\u754c\u306b\u306a\u308a\u307e\u3057\u305f\u3002<strong>\u79fb\u884c\u30ac\u30a4\u30c9\u306e\u65e2\u5b9a\u5024\u5909\u66f4<\/strong>\uff08JDBC\u306e\u578b\u30de\u30c3\u30d4\u30f3\u30b0\u542b\u3080\uff09\u3092\u8aad\u307f\u3001<strong>\u4e92\u63db\u30d5\u30e9\u30b0<\/strong>\u3067\u4e00\u6642\u7684\u306b\u56de\u907f\u3057\u3064\u3064\u3001<strong>\u8a2d\u8a08\u306e\u672c\u4e38\uff08\u578b\u30fb\u5236\u7d04\uff09<\/strong>\u3092\u6b63\u3059\u306e\u304c\u738b\u9053\u3002<\/p>\r\n<p><!-- notionvc: 6fb01ca4-4a2a-4996-aeb4-5385d067fa97 --><\/p>\n\n<h2>\u5b9f\u88c5\u6226\u7565\uff1a2025\u5e74\u306e\u30d9\u30b9\u30c8\u30d7\u30e9\u30af\u30c6\u30a3\u30b9<\/h2>\n<b>\u30b9\u30ad\u30fc\u30de\u306f\u300c\u660e\u793a\u300d\u3002\u7d71\u8a08\u306f\u300c\u66f4\u65b0\u300d\u3002\u30b3\u30b9\u30c8\u306f\u300c\u63a8\u5b9a\u3055\u305b\u308b\u300d<\/b>\r\n<ul>\r\n\t<li><strong>\u30b9\u30ad\u30fc\u30de\u660e\u793a<\/strong>\u3067\u63a8\u8ad6\u30b3\u30b9\u30c8\u3068\u578b\u30d6\u30ec\u3092\u6392\u9664\u3002<\/li>\r\n\t<li><strong>ANALYZE TABLE<\/strong>\u3092<strong>\u5b9a\u671f\u30b8\u30e7\u30d6<\/strong>\u5316\u3002<strong>\u30b3\u30b9\u30c8\u30d9\u30fc\u30b9\u6700\u9069\u5316\uff08CBO\uff09<\/strong>\u306b\u990c\u3092\u4e0e\u3048\u308b\u3002<\/li>\r\n\t<li><strong>EXPLAIN<\/strong>\u3092CI\u3067\u81ea\u52d5\u4fdd\u5b58\u3057\u3001<strong>\u56de\u5e30\u691c\u77e5<\/strong>\u3002\u30b3\u30df\u30c3\u30bf\u30fc\u89e3\u8aac\u306e\u3068\u304a\u308a\u3001<strong>\u30d7\u30e9\u30f3\u3092\u8aad\u3080\u6587\u5316<\/strong>\u304c\u5f37\u3044\u30c1\u30fc\u30e0\u306f\u901f\u3044\u3002<\/li>\r\n<\/ul>\r\n<b>\u30c7\u30fc\u30bf\u306f\u5217\u6307\u5411\uff0b\u30d1\u30fc\u30c6\u30a3\u30b7\u30e7\u30f3\u306e\u4e8c\u6bb5\u69cb\u3048<\/b>\r\n<ul>\r\n\t<li><strong>Parquet\/ORC<\/strong>\u3067\u5217\u526a\u5b9a\u3068\u30d5\u30a3\u30eb\u30bf\u4e0b\u63a8\u3057\u3092\u5f97\u308b\u3002<\/li>\r\n\t<li>\u30d1\u30fc\u30c6\u30a3\u30b7\u30e7\u30f3\u306f<strong>\u9ad8\u9078\u629e\u6027\u00d7\u904e\u5206\u5272\u3057\u306a\u3044<\/strong>\u30d0\u30e9\u30f3\u30b9\u3067\u3002<strong>AQE<\/strong>\u304c\u3042\u308b\u3068\u306f\u3044\u3048\u3001<strong>\u7269\u7406\u8a2d\u8a08\u3092\u30b5\u30dc\u3089\u306a\u3044<\/strong>\u307b\u3046\u304c\u30ea\u30bd\u30fc\u30b9\u52b9\u7387\u306f\u7d50\u5c40\u826f\u3044\u3002<\/li>\r\n<\/ul>\r\n<b>\u30ac\u30d0\u30ca\u30f3\u30b9\u306fUnity Catalog\u3092\u65e2\u5b9a\u306b<\/b>\r\n<ul>\r\n\t<li><strong>\u6a29\u9650\u30fb\u76e3\u67fb\u30fb\u30ea\u30cd\u30fc\u30b8<\/strong>\u30fb\u30c6\u30fc\u30d6\u30eb\u7ba1\u7406\u3092<strong>UC\u3067\u7d71\u5408<\/strong>\u3002<strong>\u30ec\u30ac\u30b7\u30fc\u6a29\u9650\u30e2\u30c7\u30eb\u306f\u6bb5\u968e\u7684\u5ec3\u6b62<\/strong>\u306e\u6d41\u308c\u3002<strong>Managed Table<\/strong>\u3067\u30b9\u30c8\u30ec\u30fc\u30b8\u968e\u5c64\u3068\u6700\u9069\u5316\u3092\u4e00\u4f53\u5316\u3002<\/li>\r\n<\/ul>\r\n<b>\u30a2\u30d7\u30ea\u306fSpark Connect\u3067\u758e\u7d50\u5408\u306b<\/b>\r\n<ul>\r\n\t<li><strong>\u8efd\u91cf\u30af\u30e9\u30a4\u30a2\u30f3\u30c8<\/strong>\u3067\u30de\u30a4\u30af\u30ed\u30b5\u30fc\u30d3\u30b9\u304b\u3089Spark\u3092\u547c\u3076\u3002<strong>\u8a08\u753b\u306f\u30b5\u30fc\u30d0\u3067\u6700\u9069\u5316<\/strong>\u3001\u7d50\u679c\u306f<strong>Arrow\u3067\u30b9\u30c8\u30ea\u30fc\u30e0\u8fd4\u5374<\/strong>\u3002<strong>\u8a00\u8a9e\u6a2a\u65ad<\/strong>\u30fb<strong>\u30bb\u30ad\u30e5\u30ea\u30c6\u30a3\u5883\u754c<\/strong>\u30fb<strong>\u904b\u7528\u306e\u72ec\u7acb\u6027<\/strong>\u304c\u624b\u306b\u5165\u308b\u3002<\/li>\r\n<\/ul>\r\n<p><!-- notionvc: 91024a8a-6ce4-461d-ad6f-5f3e7b34f023 --><\/p>\n\n<h2>\u307e\u3068\u3081\uff1aSQL\u3067\u300c\u6b63\u3057\u304f\u901f\u304f\u300d\u4f5c\u308b\u529b\u304c\u3001\u30c7\u30fc\u30bf\u57fa\u76e4\u306e\u5dee\u306b\u306a\u308b<\/h2>\n<p>SparkSQL\u306f\u3001<strong>\u5ba3\u8a00\u7684API\u00d7\u5b9f\u884c\u6642\u6700\u9069\u5316<\/strong>\u3068\u3044\u3046\u738b\u9053\u306e\u8a2d\u8a08\u3067\u3001ETL\u30fb\u5206\u6790\u30fbML\u524d\u51e6\u7406\u30fb\u30b9\u30c8\u30ea\u30fc\u30df\u30f3\u30b0\u3092<strong>\u4e00\u3064\u306e\u601d\u8003\u6cd5<\/strong>\u306b\u307e\u3068\u3081\u3066\u304f\u308c\u307e\u3059\u30022025\u5e74\u306e\u7126\u70b9\u306f\u3001<strong>ANSI\u65e2\u5b9a\u5316\u306b\u3088\u308b\u5b89\u5168\u6027\u306e\u5e95\u4e0a\u3052<\/strong>\u3001<strong>Spark Connect \u306b\u3088\u308b\u758e\u7d50\u5408\u306e\u958b\u767a\u4f53\u9a13<\/strong>\u3001\u305d\u3057\u3066<strong>Unity Catalog \u3067\u306e\u30ac\u30d0\u30ca\u30f3\u30b9\u4e00\u4f53\u5316<\/strong>\u3002<\/p>\r\n<p>\u3053\u306e3\u70b9\u3092\u62bc\u3055\u3048\u3001<strong>\u5217\u6307\u5411\u30fb\u30d1\u30fc\u30c6\u30a3\u30b7\u30e7\u30f3\u30fb\u7d71\u8a08\u30fbAQE<\/strong>\u3068\u3044\u3046\u57fa\u672c\u306b\u5fe0\u5b9f\u3067\u3042\u308c\u3070\u3001\u898f\u6a21\u304c\u5897\u3057\u3066\u3082<strong>\u901f\u3055\u3068\u518d\u73fe\u6027<\/strong>\u3092\u4e21\u7acb\u3067\u304d\u307e\u3059\u3002\u30cf\u30f3\u30ba\u30aa\u30f3\u3068\u4e00\u6b21\u60c5\u5831\u3067\u3001<strong>\u300c\u30d7\u30e9\u30f3\u3092\u8aad\u3080\u300d\u6587\u5316<\/strong>\u3092\u30c1\u30fc\u30e0\u306b\u6839\u3065\u304b\u305b\u308b\u3053\u3068\u304b\u3089\u59cb\u3081\u307e\u3057\u3087\u3046\u3002<\/p>\r\n<p><!-- notionvc: 1e7f45f3-e1ee-4d76-a78f-de18cdcb11d1 --><\/p>","protected":false},"excerpt":{"rendered":"<p>SparkSQL\u306e\u57fa\u672c\u6982\u5ff5\u304b\u3089\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3001\u30d7\u30ed\u30bb\u30b9\u3001\u5b9f\u8df5\u624b\u6cd5\u3001\u6700\u65b0\u30c8\u30ec\u30f3\u30c9\uff08Spark 4.0\u30fbANSI\u30e2\u30fc\u30c9\u65e2\u5b9a\u5316\u30fbSpark Connect\uff09\u307e\u3067\u3092\u3001\u30c7\u30fc\u30bf\u57fa\u76e4\u306e\u610f\u601d\u6c7a\u5b9a\u306b\u5f79\u7acb\u3064\u89b3\u70b9\u3067\u89e3\u8aac\u3057\u307e\u3059\u3002\u30e6\u30fc\u30b9\u30b1\u30fc\u30b9\u3084\u30c1 [&hellip;]<\/p>\n","protected":false},"author":22,"featured_media":7204,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"content-type":"","swell_btn_cv_data":"","footnotes":"","_wp_rev_ctl_limit":""},"categories":[1249],"tags":[1150,1149,1151,1148,811],"class_list":["post-7200","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-knowledge","tag-ansi","tag-catalyst","tag-sparkconnect","tag-sparksql","tag-811"],"_links":{"self":[{"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/posts\/7200","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/users\/22"}],"replies":[{"embeddable":true,"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/comments?post=7200"}],"version-history":[{"count":1,"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/posts\/7200\/revisions"}],"predecessor-version":[{"id":7212,"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/posts\/7200\/revisions\/7212"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/media\/7204"}],"wp:attachment":[{"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/media?parent=7200"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/categories?post=7200"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/since2020.jp\/media\/wp-json\/wp\/v2\/tags?post=7200"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}