Github user davies commented on a diff in the pull request:
https://github.com/apache/spark/pull/10589#discussion_r48900041
--- Diff:
sql/core/src/test/scala/org/apache/spark/sql/execution/datasources/parquet/ParquetReadBenchmark.scala
---
@@ -0,0 +1,157 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.spark.sql.execution.datasources.parquet
+
+import java.io.File
+
+import scala.collection.JavaConverters._
+import scala.util.Try
+
+import org.apache.spark.sql.{SQLConf, SQLContext}
+import org.apache.spark.util.Utils
+import org.apache.spark.{Benchmark, SparkConf, SparkContext}
+
+/**
+ * Benchmark to measure parquet read performance.
+ */
+object ParquetReadBenchmark {
+ val conf = new SparkConf()
+ conf.set("spark.sql.parquet.compression.codec", "snappy")
+ val sc = new SparkContext("local[1]", "test-sql-context", conf)
+ val sqlContext = new SQLContext(sc)
+
+ def withTempPath(f: File => Unit): Unit = {
+ val path = Utils.createTempDir()
+ path.delete()
+ try f(path) finally Utils.deleteRecursively(path)
+ }
+
+ def withTempTable(tableNames: String*)(f: => Unit): Unit = {
+ try f finally tableNames.foreach(sqlContext.dropTempTable)
+ }
+
+ def withSQLConf(pairs: (String, String)*)(f: => Unit): Unit = {
+ val (keys, values) = pairs.unzip
+ val currentValues = keys.map(key =>
Try(sqlContext.conf.getConfString(key)).toOption)
+ (keys, values).zipped.foreach(sqlContext.conf.setConfString)
+ try f finally {
+ keys.zip(currentValues).foreach {
+ case (key, Some(value)) => sqlContext.conf.setConfString(key,
value)
+ case (key, None) => sqlContext.conf.unsetConf(key)
+ }
+ }
+ }
+
+ def intScanBenchmark(values: Int): Unit = {
+ withTempPath { dir =>
+ sqlContext.range(values).write.parquet(dir.getCanonicalPath)
+ withTempTable("tempTable") {
+
sqlContext.read.parquet(dir.getCanonicalPath).registerTempTable("tempTable")
+ val benchmark = new Benchmark("Single Int Column Scan", values)
+
+ benchmark.addCase("SQL", iter => {
+ sqlContext.sql("select sum(id) from tempTable").collect()
+ })
+
+ benchmark.addCase("SQL Parquet MR", iter => {
+ withSQLConf(SQLConf.PARQUET_UNSAFE_ROW_RECORD_READER_ENABLED.key
-> "false") {
+ sqlContext.sql("select sum(id) from tempTable").collect()
+ }
+ })
+
+ val files =
SpecificParquetRecordReaderBase.listDirectory(dir).toArray
+ benchmark.addCase("ParquetReader", num => {
+ var sum = 0L
+ files.map(_.asInstanceOf[String]).foreach { p =>
+ val reader = new UnsafeRowParquetRecordReader
+ reader.initialize(p, ("id" :: Nil).asJava)
+
+ while (reader.nextKeyValue()) {
+ val record = reader.getCurrentValue
+ if (!record.isNullAt(0)) sum += record.getInt(0)
+ }
+ reader.close()
+ }})
+
+ /*
+ Intel(R) Core(TM) i7-4870HQ CPU @ 2.50GHz
+ Single Int Column Scan: Avg Time(ms) Avg Rate(M/s)
Relative Rate
+
-------------------------------------------------------------------------
+ SQL 1910.0 13.72
1.00 X
--- End diff --
This should be `SQL with ParquetReader`
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
