Repository: spark Updated Branches: refs/heads/master 5330c192b -> 62551ccee
[SPARK-25492][TEST] Refactor WideSchemaBenchmark to use main method ## What changes were proposed in this pull request? Refactor `WideSchemaBenchmark` to use main method. 1. use `spark-submit`: ```console bin/spark-submit --class org.apache.spark.sql.execution.benchmark.WideSchemaBenchmark --jars ./core/target/spark-core_2.11-3.0.0-SNAPSHOT-tests.jar ./sql/core/target/spark-sql_2.11-3.0.0-SNAPSHOT-tests.jar ``` 2. Generate benchmark result: ```console SPARK_GENERATE_BENCHMARK_FILES=1 build/sbt "sql/test:runMain org.apache.spark.sql.execution.benchmark.WideSchemaBenchmark" ``` ## How was this patch tested? manual tests Closes #22501 from wangyum/SPARK-25492. Lead-authored-by: Yuming Wang <[email protected]> Co-authored-by: Dongjoon Hyun <[email protected]> Signed-off-by: Dongjoon Hyun <[email protected]> Project: http://git-wip-us.apache.org/repos/asf/spark/repo Commit: http://git-wip-us.apache.org/repos/asf/spark/commit/62551cce Tree: http://git-wip-us.apache.org/repos/asf/spark/tree/62551cce Diff: http://git-wip-us.apache.org/repos/asf/spark/diff/62551cce Branch: refs/heads/master Commit: 62551cceebf6aca8b6bd8164cd2ed85564726f6c Parents: 5330c19 Author: Yuming Wang <[email protected]> Authored: Sat Oct 20 17:31:13 2018 -0700 Committer: Dongjoon Hyun <[email protected]> Committed: Sat Oct 20 17:31:13 2018 -0700 ---------------------------------------------------------------------- .../benchmarks/WideSchemaBenchmark-results.txt | 216 +++++++++++-------- .../benchmark/WideSchemaBenchmark.scala | 134 +++++++----- 2 files changed, 197 insertions(+), 153 deletions(-) ---------------------------------------------------------------------- http://git-wip-us.apache.org/repos/asf/spark/blob/62551cce/sql/core/benchmarks/WideSchemaBenchmark-results.txt ---------------------------------------------------------------------- diff --git a/sql/core/benchmarks/WideSchemaBenchmark-results.txt b/sql/core/benchmarks/WideSchemaBenchmark-results.txt index 0b9f791..6347a6a 100644 --- a/sql/core/benchmarks/WideSchemaBenchmark-results.txt +++ b/sql/core/benchmarks/WideSchemaBenchmark-results.txt @@ -1,117 +1,145 @@ -Java HotSpot(TM) 64-Bit Server VM 1.8.0_92-b14 on Mac OS X 10.11.6 -Intel(R) Core(TM) i7-4980HQ CPU @ 2.80GHz +================================================================================================ +parsing large select expressions +================================================================================================ +OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64 +Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz parsing large select: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------ -1 select expressions 2 / 4 0.0 2050147.0 1.0X -100 select expressions 6 / 7 0.0 6123412.0 0.3X -2500 select expressions 135 / 141 0.0 134623148.0 0.0X +1 select expressions 6 / 13 0.0 5997373.0 1.0X +100 select expressions 7 / 10 0.0 7204596.0 0.8X +2500 select expressions 103 / 107 0.0 102962705.0 0.1X -Java HotSpot(TM) 64-Bit Server VM 1.8.0_92-b14 on Mac OS X 10.11.6 -Intel(R) Core(TM) i7-4980HQ CPU @ 2.80GHz +================================================================================================ +many column field read and write +================================================================================================ + +OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64 +Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz many column field r/w: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------ -1 cols x 100000 rows (read in-mem) 16 / 18 6.3 158.6 1.0X -1 cols x 100000 rows (exec in-mem) 17 / 19 6.0 166.7 1.0X -1 cols x 100000 rows (read parquet) 24 / 26 4.3 235.1 0.7X -1 cols x 100000 rows (write parquet) 81 / 85 1.2 811.3 0.2X -100 cols x 1000 rows (read in-mem) 17 / 19 6.0 166.2 1.0X -100 cols x 1000 rows (exec in-mem) 25 / 27 4.0 249.2 0.6X -100 cols x 1000 rows (read parquet) 23 / 25 4.4 226.0 0.7X -100 cols x 1000 rows (write parquet) 83 / 87 1.2 831.0 0.2X -2500 cols x 40 rows (read in-mem) 132 / 137 0.8 1322.9 0.1X -2500 cols x 40 rows (exec in-mem) 326 / 330 0.3 3260.6 0.0X -2500 cols x 40 rows (read parquet) 831 / 839 0.1 8305.8 0.0X -2500 cols x 40 rows (write parquet) 237 / 245 0.4 2372.6 0.1X - -Java HotSpot(TM) 64-Bit Server VM 1.8.0_92-b14 on Mac OS X 10.11.6 -Intel(R) Core(TM) i7-4980HQ CPU @ 2.80GHz +1 cols x 100000 rows (read in-mem) 40 / 51 2.5 396.5 1.0X +1 cols x 100000 rows (exec in-mem) 41 / 48 2.4 414.4 1.0X +1 cols x 100000 rows (read parquet) 61 / 70 1.6 610.2 0.6X +1 cols x 100000 rows (write parquet) 209 / 233 0.5 2086.1 0.2X +100 cols x 1000 rows (read in-mem) 43 / 49 2.3 433.8 0.9X +100 cols x 1000 rows (exec in-mem) 57 / 66 1.8 568.4 0.7X +100 cols x 1000 rows (read parquet) 60 / 66 1.7 599.0 0.7X +100 cols x 1000 rows (write parquet) 212 / 224 0.5 2120.6 0.2X +2500 cols x 40 rows (read in-mem) 268 / 275 0.4 2676.5 0.1X +2500 cols x 40 rows (exec in-mem) 494 / 504 0.2 4936.9 0.1X +2500 cols x 40 rows (read parquet) 132 / 139 0.8 1319.7 0.3X +2500 cols x 40 rows (write parquet) 371 / 381 0.3 3710.1 0.1X + + +================================================================================================ +wide shallowly nested struct field read and write +================================================================================================ +OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64 +Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz wide shallowly nested struct field r/w: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------ -1 wide x 100000 rows (read in-mem) 15 / 17 6.6 151.0 1.0X -1 wide x 100000 rows (exec in-mem) 20 / 22 5.1 196.6 0.8X -1 wide x 100000 rows (read parquet) 59 / 63 1.7 592.8 0.3X -1 wide x 100000 rows (write parquet) 81 / 87 1.2 814.6 0.2X -100 wide x 1000 rows (read in-mem) 21 / 25 4.8 208.7 0.7X -100 wide x 1000 rows (exec in-mem) 72 / 81 1.4 718.5 0.2X -100 wide x 1000 rows (read parquet) 75 / 85 1.3 752.6 0.2X -100 wide x 1000 rows (write parquet) 88 / 95 1.1 876.7 0.2X -2500 wide x 40 rows (read in-mem) 28 / 34 3.5 282.2 0.5X -2500 wide x 40 rows (exec in-mem) 1269 / 1284 0.1 12688.1 0.0X -2500 wide x 40 rows (read parquet) 549 / 578 0.2 5493.4 0.0X -2500 wide x 40 rows (write parquet) 96 / 104 1.0 959.1 0.2X - -Java HotSpot(TM) 64-Bit Server VM 1.8.0_92-b14 on Mac OS X 10.11.6 -Intel(R) Core(TM) i7-4980HQ CPU @ 2.80GHz +1 wide x 100000 rows (read in-mem) 37 / 43 2.7 373.6 1.0X +1 wide x 100000 rows (exec in-mem) 47 / 54 2.1 472.7 0.8X +1 wide x 100000 rows (read parquet) 132 / 145 0.8 1316.5 0.3X +1 wide x 100000 rows (write parquet) 205 / 232 0.5 2046.3 0.2X +100 wide x 1000 rows (read in-mem) 68 / 79 1.5 676.3 0.6X +100 wide x 1000 rows (exec in-mem) 88 / 97 1.1 882.2 0.4X +100 wide x 1000 rows (read parquet) 197 / 234 0.5 1971.8 0.2X +100 wide x 1000 rows (write parquet) 236 / 249 0.4 2359.6 0.2X +2500 wide x 40 rows (read in-mem) 77 / 85 1.3 768.0 0.5X +2500 wide x 40 rows (exec in-mem) 386 / 393 0.3 3855.2 0.1X +2500 wide x 40 rows (read parquet) 1741 / 1765 0.1 17408.3 0.0X +2500 wide x 40 rows (write parquet) 243 / 256 0.4 2425.2 0.2X + +================================================================================================ +deeply nested struct field read and write +================================================================================================ + +OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64 +Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz deeply nested struct field r/w: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------ -1 deep x 100000 rows (read in-mem) 14 / 16 7.0 143.8 1.0X -1 deep x 100000 rows (exec in-mem) 17 / 19 5.9 169.7 0.8X -1 deep x 100000 rows (read parquet) 33 / 35 3.1 327.0 0.4X -1 deep x 100000 rows (write parquet) 79 / 84 1.3 786.9 0.2X -100 deep x 1000 rows (read in-mem) 21 / 24 4.7 211.3 0.7X -100 deep x 1000 rows (exec in-mem) 221 / 235 0.5 2214.5 0.1X -100 deep x 1000 rows (read parquet) 1928 / 1952 0.1 19277.1 0.0X -100 deep x 1000 rows (write parquet) 91 / 96 1.1 909.5 0.2X -250 deep x 400 rows (read in-mem) 57 / 61 1.8 567.1 0.3X -250 deep x 400 rows (exec in-mem) 1329 / 1385 0.1 13291.8 0.0X -250 deep x 400 rows (read parquet) 36563 / 36750 0.0 365630.2 0.0X -250 deep x 400 rows (write parquet) 126 / 130 0.8 1262.0 0.1X - -Java HotSpot(TM) 64-Bit Server VM 1.8.0_92-b14 on Mac OS X 10.11.6 -Intel(R) Core(TM) i7-4980HQ CPU @ 2.80GHz +1 deep x 100000 rows (read in-mem) 35 / 42 2.9 350.2 1.0X +1 deep x 100000 rows (exec in-mem) 40 / 43 2.5 399.5 0.9X +1 deep x 100000 rows (read parquet) 69 / 73 1.4 691.6 0.5X +1 deep x 100000 rows (write parquet) 203 / 224 0.5 2025.9 0.2X +100 deep x 1000 rows (read in-mem) 70 / 75 1.4 703.7 0.5X +100 deep x 1000 rows (exec in-mem) 654 / 684 0.2 6539.9 0.1X +100 deep x 1000 rows (read parquet) 10503 / 10550 0.0 105030.5 0.0X +100 deep x 1000 rows (write parquet) 235 / 243 0.4 2353.2 0.1X +250 deep x 400 rows (read in-mem) 249 / 259 0.4 2492.6 0.1X +250 deep x 400 rows (exec in-mem) 3842 / 3854 0.0 38424.8 0.0X +250 deep x 400 rows (read parquet) 153080 / 153444 0.0 1530796.1 0.0X +250 deep x 400 rows (write parquet) 434 / 441 0.2 4344.6 0.1X + +================================================================================================ +bushy struct field read and write +================================================================================================ + +OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64 +Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz bushy struct field r/w: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------ -1 x 1 deep x 100000 rows (read in-mem) 13 / 15 7.8 127.7 1.0X -1 x 1 deep x 100000 rows (exec in-mem) 15 / 17 6.6 151.5 0.8X -1 x 1 deep x 100000 rows (read parquet) 20 / 23 5.0 198.3 0.6X -1 x 1 deep x 100000 rows (write parquet) 77 / 82 1.3 770.4 0.2X -128 x 8 deep x 1000 rows (read in-mem) 12 / 14 8.2 122.5 1.0X -128 x 8 deep x 1000 rows (exec in-mem) 124 / 140 0.8 1241.2 0.1X -128 x 8 deep x 1000 rows (read parquet) 69 / 74 1.4 693.9 0.2X -128 x 8 deep x 1000 rows (write parquet) 78 / 83 1.3 777.7 0.2X -1024 x 11 deep x 100 rows (read in-mem) 25 / 29 4.1 246.1 0.5X -1024 x 11 deep x 100 rows (exec in-mem) 1197 / 1223 0.1 11974.6 0.0X -1024 x 11 deep x 100 rows (read parquet) 426 / 433 0.2 4263.7 0.0X -1024 x 11 deep x 100 rows (write parquet) 91 / 98 1.1 913.5 0.1X - -Java HotSpot(TM) 64-Bit Server VM 1.8.0_92-b14 on Mac OS X 10.11.6 -Intel(R) Core(TM) i7-4980HQ CPU @ 2.80GHz +1 x 1 deep x 100000 rows (read in-mem) 37 / 42 2.7 370.2 1.0X +1 x 1 deep x 100000 rows (exec in-mem) 43 / 47 2.4 425.3 0.9X +1 x 1 deep x 100000 rows (read parquet) 48 / 51 2.1 478.7 0.8X +1 x 1 deep x 100000 rows (write parquet) 204 / 215 0.5 2042.0 0.2X +128 x 8 deep x 1000 rows (read in-mem) 32 / 37 3.1 318.6 1.2X +128 x 8 deep x 1000 rows (exec in-mem) 91 / 96 1.1 906.6 0.4X +128 x 8 deep x 1000 rows (read parquet) 351 / 379 0.3 3510.3 0.1X +128 x 8 deep x 1000 rows (write parquet) 199 / 203 0.5 1988.3 0.2X +1024 x 11 deep x 100 rows (read in-mem) 73 / 76 1.4 730.4 0.5X +1024 x 11 deep x 100 rows (exec in-mem) 327 / 334 0.3 3267.2 0.1X +1024 x 11 deep x 100 rows (read parquet) 2063 / 2078 0.0 20629.2 0.0X +1024 x 11 deep x 100 rows (write parquet) 248 / 266 0.4 2475.1 0.1X + + +================================================================================================ +wide array field read and write +================================================================================================ +OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64 +Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz wide array field r/w: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------ -1 wide x 100000 rows (read in-mem) 14 / 16 7.0 143.2 1.0X -1 wide x 100000 rows (exec in-mem) 17 / 19 5.9 170.9 0.8X -1 wide x 100000 rows (read parquet) 43 / 46 2.3 434.1 0.3X -1 wide x 100000 rows (write parquet) 78 / 83 1.3 777.6 0.2X -100 wide x 1000 rows (read in-mem) 11 / 13 9.0 111.5 1.3X -100 wide x 1000 rows (exec in-mem) 13 / 15 7.8 128.3 1.1X -100 wide x 1000 rows (read parquet) 24 / 27 4.1 245.0 0.6X -100 wide x 1000 rows (write parquet) 74 / 80 1.4 740.5 0.2X -2500 wide x 40 rows (read in-mem) 11 / 13 9.1 109.5 1.3X -2500 wide x 40 rows (exec in-mem) 13 / 15 7.7 129.4 1.1X -2500 wide x 40 rows (read parquet) 24 / 26 4.1 241.3 0.6X -2500 wide x 40 rows (write parquet) 75 / 81 1.3 751.8 0.2X - -Java HotSpot(TM) 64-Bit Server VM 1.8.0_92-b14 on Mac OS X 10.11.6 -Intel(R) Core(TM) i7-4980HQ CPU @ 2.80GHz +1 wide x 100000 rows (read in-mem) 33 / 38 3.0 328.4 1.0X +1 wide x 100000 rows (exec in-mem) 40 / 44 2.5 402.7 0.8X +1 wide x 100000 rows (read parquet) 83 / 91 1.2 826.6 0.4X +1 wide x 100000 rows (write parquet) 204 / 218 0.5 2039.1 0.2X +100 wide x 1000 rows (read in-mem) 28 / 31 3.6 277.2 1.2X +100 wide x 1000 rows (exec in-mem) 34 / 37 2.9 343.2 1.0X +100 wide x 1000 rows (read parquet) 56 / 61 1.8 556.4 0.6X +100 wide x 1000 rows (write parquet) 202 / 206 0.5 2017.3 0.2X +2500 wide x 40 rows (read in-mem) 29 / 30 3.5 286.4 1.1X +2500 wide x 40 rows (exec in-mem) 33 / 39 3.0 330.2 1.0X +2500 wide x 40 rows (read parquet) 54 / 66 1.8 544.0 0.6X +2500 wide x 40 rows (write parquet) 196 / 208 0.5 1959.2 0.2X + +================================================================================================ +wide map field read and write +================================================================================================ + +OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64 +Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz wide map field r/w: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------ -1 wide x 100000 rows (read in-mem) 16 / 18 6.2 162.6 1.0X -1 wide x 100000 rows (exec in-mem) 21 / 23 4.8 208.2 0.8X -1 wide x 100000 rows (read parquet) 54 / 59 1.8 543.6 0.3X -1 wide x 100000 rows (write parquet) 80 / 86 1.2 804.5 0.2X -100 wide x 1000 rows (read in-mem) 11 / 13 8.7 114.5 1.4X -100 wide x 1000 rows (exec in-mem) 14 / 16 7.0 143.5 1.1X -100 wide x 1000 rows (read parquet) 30 / 32 3.3 300.4 0.5X -100 wide x 1000 rows (write parquet) 75 / 80 1.3 749.9 0.2X -2500 wide x 40 rows (read in-mem) 13 / 15 7.8 128.1 1.3X -2500 wide x 40 rows (exec in-mem) 15 / 18 6.5 153.6 1.1X -2500 wide x 40 rows (read parquet) 30 / 33 3.3 304.4 0.5X -2500 wide x 40 rows (write parquet) 77 / 83 1.3 768.5 0.2X +1 wide x 100000 rows (read in-mem) 31 / 34 3.3 305.7 1.0X +1 wide x 100000 rows (exec in-mem) 39 / 44 2.6 390.0 0.8X +1 wide x 100000 rows (read parquet) 125 / 132 0.8 1250.5 0.2X +1 wide x 100000 rows (write parquet) 198 / 213 0.5 1979.9 0.2X +100 wide x 1000 rows (read in-mem) 21 / 23 4.7 212.7 1.4X +100 wide x 1000 rows (exec in-mem) 28 / 32 3.5 283.3 1.1X +100 wide x 1000 rows (read parquet) 68 / 73 1.5 683.0 0.4X +100 wide x 1000 rows (write parquet) 188 / 206 0.5 1882.1 0.2X +2500 wide x 40 rows (read in-mem) 25 / 28 4.0 252.2 1.2X +2500 wide x 40 rows (exec in-mem) 32 / 34 3.1 318.5 1.0X +2500 wide x 40 rows (read parquet) 69 / 73 1.4 691.5 0.4X +2500 wide x 40 rows (write parquet) 193 / 202 0.5 1932.8 0.2X + http://git-wip-us.apache.org/repos/asf/spark/blob/62551cce/sql/core/src/test/scala/org/apache/spark/sql/execution/benchmark/WideSchemaBenchmark.scala ---------------------------------------------------------------------- diff --git a/sql/core/src/test/scala/org/apache/spark/sql/execution/benchmark/WideSchemaBenchmark.scala b/sql/core/src/test/scala/org/apache/spark/sql/execution/benchmark/WideSchemaBenchmark.scala index 81017a6..1246619 100644 --- a/sql/core/src/test/scala/org/apache/spark/sql/execution/benchmark/WideSchemaBenchmark.scala +++ b/sql/core/src/test/scala/org/apache/spark/sql/execution/benchmark/WideSchemaBenchmark.scala @@ -15,59 +15,36 @@ * limitations under the License. */ -package org.apache.spark.sql +package org.apache.spark.sql.execution.benchmark -import java.io.{File, FileOutputStream, OutputStream} +import java.io.File -import org.scalatest.BeforeAndAfterEach - -import org.apache.spark.SparkFunSuite import org.apache.spark.benchmark.Benchmark -import org.apache.spark.sql.functions._ +import org.apache.spark.sql.DataFrame import org.apache.spark.util.Utils /** * Benchmark for performance with very wide and nested DataFrames. - * To run this: - * build/sbt "sql/test-only *WideSchemaBenchmark" - * - * Results will be written to "sql/core/benchmarks/WideSchemaBenchmark-results.txt". + * To run this benchmark: + * {{{ + * 1. without sbt: + * bin/spark-submit --class <this class> --jars <spark core test jar> <spark sql test jar> + * 2. build/sbt "sql/test:runMain <this class>" + * 3. generate result: SPARK_GENERATE_BENCHMARK_FILES=1 build/sbt "sql/test:runMain <this class>" + * Results will be written to "benchmarks/WideSchemaBenchmark-results.txt". + * }}} */ -class WideSchemaBenchmark extends SparkFunSuite with BeforeAndAfterEach { +object WideSchemaBenchmark extends SqlBasedBenchmark { private val scaleFactor = 100000 private val widthsToTest = Seq(1, 100, 2500) private val depthsToTest = Seq(1, 100, 250) assert(scaleFactor > widthsToTest.max) - private lazy val sparkSession = SparkSession.builder - .master("local[1]") - .appName("microbenchmark") - .getOrCreate() - - import sparkSession.implicits._ + import spark.implicits._ private var tmpFiles: List[File] = Nil - private var out: OutputStream = null - - override def beforeAll() { - super.beforeAll() - out = new FileOutputStream(new File("benchmarks/WideSchemaBenchmark-results.txt")) - } - override def afterAll() { - try { - out.close() - } finally { - super.afterAll() - } - } - - override def afterEach() { - super.afterEach() - for (tmpFile <- tmpFiles) { - Utils.deleteRecursively(tmpFile) - } - } + private def deleteTmpFiles(): Unit = tmpFiles.foreach(Utils.deleteRecursively) /** * Writes the given DataFrame to parquet at a temporary location, and returns a DataFrame @@ -79,7 +56,7 @@ class WideSchemaBenchmark extends SparkFunSuite with BeforeAndAfterEach { tmpFile.delete() df.write.parquet(tmpFile.getAbsolutePath) assert(tmpFile.isDirectory()) - sparkSession.read.parquet(tmpFile.getAbsolutePath) + spark.read.parquet(tmpFile.getAbsolutePath) } /** @@ -105,33 +82,33 @@ class WideSchemaBenchmark extends SparkFunSuite with BeforeAndAfterEach { } } - ignore("parsing large select expressions") { - val benchmark = new Benchmark("parsing large select", 1, output = Some(out)) + def parsingLargeSelectExpressions(): Unit = { + val benchmark = new Benchmark("parsing large select", 1, output = output) for (width <- widthsToTest) { val selectExpr = (1 to width).map(i => s"id as a_$i") benchmark.addCase(s"$width select expressions") { iter => - sparkSession.range(1).toDF.selectExpr(selectExpr: _*) + spark.range(1).toDF.selectExpr(selectExpr: _*) } } benchmark.run() } - ignore("many column field read and write") { - val benchmark = new Benchmark("many column field r/w", scaleFactor, output = Some(out)) + def manyColumnFieldReadAndWrite(): Unit = { + val benchmark = new Benchmark("many column field r/w", scaleFactor, output = output) for (width <- widthsToTest) { // normalize by width to keep constant data size val numRows = scaleFactor / width val selectExpr = (1 to width).map(i => s"id as a_$i") - val df = sparkSession.range(numRows).toDF.selectExpr(selectExpr: _*).cache() + val df = spark.range(numRows).toDF.selectExpr(selectExpr: _*).cache() df.count() // force caching addCases(benchmark, df, s"$width cols x $numRows rows", "a_1") } benchmark.run() } - ignore("wide shallowly nested struct field read and write") { + def wideShallowlyNestedStructFieldReadAndWrite(): Unit = { val benchmark = new Benchmark( - "wide shallowly nested struct field r/w", scaleFactor, output = Some(out)) + "wide shallowly nested struct field r/w", scaleFactor, output = output) for (width <- widthsToTest) { val numRows = scaleFactor / width var datum: String = "{" @@ -144,15 +121,15 @@ class WideSchemaBenchmark extends SparkFunSuite with BeforeAndAfterEach { } datum += "}" datum = s"""{"a": {"b": {"c": $datum, "d": $datum}, "e": $datum}}""" - val df = sparkSession.read.json(sparkSession.range(numRows).map(_ => datum)).cache() + val df = spark.read.json(spark.range(numRows).map(_ => datum)).cache() df.count() // force caching addCases(benchmark, df, s"$width wide x $numRows rows", "a.b.c.value_1") } benchmark.run() } - ignore("deeply nested struct field read and write") { - val benchmark = new Benchmark("deeply nested struct field r/w", scaleFactor, output = Some(out)) + def deeplyNestedStructFieldReadAndWrite(): Unit = { + val benchmark = new Benchmark("deeply nested struct field r/w", scaleFactor, output = output) for (depth <- depthsToTest) { val numRows = scaleFactor / depth var datum: String = "{\"value\": 1}" @@ -161,15 +138,15 @@ class WideSchemaBenchmark extends SparkFunSuite with BeforeAndAfterEach { datum = "{\"value\": " + datum + "}" selector = selector + ".value" } - val df = sparkSession.read.json(sparkSession.range(numRows).map(_ => datum)).cache() + val df = spark.read.json(spark.range(numRows).map(_ => datum)).cache() df.count() // force caching addCases(benchmark, df, s"$depth deep x $numRows rows", selector) } benchmark.run() } - ignore("bushy struct field read and write") { - val benchmark = new Benchmark("bushy struct field r/w", scaleFactor, output = Some(out)) + def bushyStructFieldReadAndWrite(): Unit = { + val benchmark = new Benchmark("bushy struct field r/w", scaleFactor, output = output) for (width <- Seq(1, 100, 1000)) { val numRows = scaleFactor / width var numNodes = 1 @@ -184,15 +161,16 @@ class WideSchemaBenchmark extends SparkFunSuite with BeforeAndAfterEach { } // TODO(ekl) seems like the json parsing is actually the majority of the time, perhaps // we should benchmark that too separately. - val df = sparkSession.read.json(sparkSession.range(numRows).map(_ => datum)).cache() + val df = spark.read.json(spark.range(numRows).map(_ => datum)).cache() df.count() // force caching addCases(benchmark, df, s"$numNodes x $depth deep x $numRows rows", selector) } benchmark.run() } - ignore("wide array field read and write") { - val benchmark = new Benchmark("wide array field r/w", scaleFactor, output = Some(out)) + + def wideArrayFieldReadAndWrite(): Unit = { + val benchmark = new Benchmark("wide array field r/w", scaleFactor, output = output) for (width <- widthsToTest) { val numRows = scaleFactor / width var datum: String = "{\"value\": [" @@ -204,22 +182,60 @@ class WideSchemaBenchmark extends SparkFunSuite with BeforeAndAfterEach { } } datum += "]}" - val df = sparkSession.read.json(sparkSession.range(numRows).map(_ => datum)).cache() + val df = spark.read.json(spark.range(numRows).map(_ => datum)).cache() df.count() // force caching addCases(benchmark, df, s"$width wide x $numRows rows", "value[0]") } benchmark.run() } - ignore("wide map field read and write") { - val benchmark = new Benchmark("wide map field r/w", scaleFactor, output = Some(out)) + def wideMapFieldReadAndWrite(): Unit = { + val benchmark = new Benchmark("wide map field r/w", scaleFactor, output = output) for (width <- widthsToTest) { val numRows = scaleFactor / width val datum = Tuple1((1 to width).map(i => ("value_" + i -> 1)).toMap) - val df = sparkSession.range(numRows).map(_ => datum).toDF.cache() + val df = spark.range(numRows).map(_ => datum).toDF.cache() df.count() // force caching addCases(benchmark, df, s"$width wide x $numRows rows", "_1[\"value_1\"]") } benchmark.run() } + + def runBenchmarkWithDeleteTmpFiles(benchmarkName: String)(func: => Any): Unit = { + runBenchmark(benchmarkName) { + func + } + deleteTmpFiles() + } + + override def runBenchmarkSuite(): Unit = { + + runBenchmarkWithDeleteTmpFiles("parsing large select expressions") { + parsingLargeSelectExpressions() + } + + runBenchmarkWithDeleteTmpFiles("many column field read and write") { + manyColumnFieldReadAndWrite() + } + + runBenchmarkWithDeleteTmpFiles("wide shallowly nested struct field read and write") { + wideShallowlyNestedStructFieldReadAndWrite() + } + + runBenchmarkWithDeleteTmpFiles("deeply nested struct field read and write") { + deeplyNestedStructFieldReadAndWrite() + } + + runBenchmarkWithDeleteTmpFiles("bushy struct field read and write") { + bushyStructFieldReadAndWrite() + } + + runBenchmarkWithDeleteTmpFiles("wide array field read and write") { + wideArrayFieldReadAndWrite() + } + + runBenchmarkWithDeleteTmpFiles("wide map field read and write") { + wideMapFieldReadAndWrite() + } + } } --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
