From d60b09bb60cff106fa0acddebf35714503b20f03 Mon Sep 17 00:00:00 2001
From: Zongheng Yang <zongheng.y@gmail.com>
Date: Mon, 14 Jul 2014 13:22:24 -0700
Subject: [PATCH] [SPARK-2443][SQL] Fix slow read from partitioned tables
This fix obtains a comparable performance boost as [PR #1390](https://github.com/apache/spark/pull/1390) by moving an array update and deserializer initialization out of a potentially very long loop. Suggested by yhuai. The below results are updated for this fix.
## Benchmarks
Generated a local text file with 10M rows of simple key-value pairs. The data is loaded as a table through Hive. Results are obtained on my local machine using hive/console.
Without the fix:
Type | Non-partitioned | Partitioned (1 part)
------------ | ------------ | -------------
First run | 9.52s end-to-end (1.64s Spark job) | 36.6s (28.3s)
Stablized runs | 1.21s (1.18s) | 27.6s (27.5s)
With this fix:
Type | Non-partitioned | Partitioned (1 part)
------------ | ------------ | -------------
First run | 9.57s (1.46s) | 11.0s (1.69s)
Stablized runs | 1.13s (1.10s) | 1.23s (1.19s)
Author: Zongheng Yang <zongheng.y@gmail.com>
Closes #1408 from concretevitamin/slow-read-2 and squashes the following commits:
d86e437 [Zongheng Yang] Move update & initialization out of potentially long loop.
---
.../scala/org/apache/spark/sql/hive/TableReader.scala | 10 +++++++---
1 file changed, 7 insertions(+), 3 deletions(-)
diff --git a/sql/hive/src/main/scala/org/apache/spark/sql/hive/TableReader.scala b/sql/hive/src/main/scala/org/apache/spark/sql/hive/TableReader.scala
index 8cfde46186..c3942578d6 100644
--- a/sql/hive/src/main/scala/org/apache/spark/sql/hive/TableReader.scala
+++ b/sql/hive/src/main/scala/org/apache/spark/sql/hive/TableReader.scala
@@ -164,13 +164,17 @@ class HadoopTableReader(@transient _tableDesc: TableDesc, @transient sc: HiveCon
hivePartitionRDD.mapPartitions { iter =>
val hconf = broadcastedHiveConf.value.value
val rowWithPartArr = new Array[Object](2)
+
+ // The update and deserializer initialization are intentionally
+ // kept out of the below iter.map loop to save performance.
+ rowWithPartArr.update(1, partValues)
+ val deserializer = localDeserializer.newInstance()
+ deserializer.initialize(hconf, partProps)
+
// Map each tuple to a row object
iter.map { value =>
- val deserializer = localDeserializer.newInstance()
- deserializer.initialize(hconf, partProps)
val deserializedRow = deserializer.deserialize(value)
rowWithPartArr.update(0, deserializedRow)
- rowWithPartArr.update(1, partValues)
rowWithPartArr.asInstanceOf[Object]
}
}
--
GitLab