You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Cheng Lian (JIRA)" <ji...@apache.org> on 2016/11/02 21:12:58 UTC
[jira] [Updated] (SPARK-17972) Query planning slows down
dramatically for large query plans even when sub-trees are cached
[ https://issues.apache.org/jira/browse/SPARK-17972?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Cheng Lian updated SPARK-17972:
-------------------------------
Description:
The following Spark shell snippet creates a series of query plans that grow exponentially. The {{i}}-th plan is created using 4 *cached* copies of the {{i - 1}}-th plan.
{code}
(0 until 6).foldLeft(Seq(1, 2, 3).toDS) { (plan, iteration) =>
val start = System.currentTimeMillis()
val result = plan.join(plan, "value").join(plan, "value").join(plan, "value").join(plan, "value")
result.cache()
System.out.println(s"Iteration $iteration takes time ${System.currentTimeMillis() - start} ms")
result.as[Int]
}
{code}
We can see that although all plans are cached, the query planning time still grows exponentially and quickly becomes unbearable.
{noformat}
Iteration 0 takes time 9 ms
Iteration 1 takes time 19 ms
Iteration 2 takes time 61 ms
Iteration 3 takes time 219 ms
Iteration 4 takes time 830 ms
Iteration 5 takes time 4080 ms
{noformat}
Similar scenarios can be found in iterative ML code and significantly affects usability.
This issue can be fixed by introducing a {{checkpoint()}} method for {{Dataset}} that truncates both the query plan and the lineage of the underlying RDD.
was:
The following Spark shell snippet creates a series of query plans that grow exponentially. The {{i}}-th plan is created using 4 *cached* copies of the {{i - 1}}-th plan.
{code}
(0 until 6).foldLeft(Seq(1, 2, 3).toDS) { (plan, iteration) =>
val start = System.currentTimeMillis()
val result = plan.join(plan, "value").join(plan, "value").join(plan, "value").join(plan, "value")
result.cache()
System.out.println(s"Iteration $iteration takes time ${System.currentTimeMillis() - start} ms")
result.as[Int]
}
{code}
We can see that although all plans are cached, the query planning time still grows exponentially and quickly becomes unbearable.
{noformat}
Iteration 0 takes time 9 ms
Iteration 1 takes time 19 ms
Iteration 2 takes time 61 ms
Iteration 3 takes time 219 ms
Iteration 4 takes time 830 ms
Iteration 5 takes time 4080 ms
{noformat}
Similar scenarios can be found in iterative ML code and significantly affects usability.
> Query planning slows down dramatically for large query plans even when sub-trees are cached
> -------------------------------------------------------------------------------------------
>
> Key: SPARK-17972
> URL: https://issues.apache.org/jira/browse/SPARK-17972
> Project: Spark
> Issue Type: Bug
> Components: SQL
> Affects Versions: 1.6.2, 2.0.1
> Reporter: Cheng Lian
> Assignee: Cheng Lian
> Fix For: 2.1.0
>
>
> The following Spark shell snippet creates a series of query plans that grow exponentially. The {{i}}-th plan is created using 4 *cached* copies of the {{i - 1}}-th plan.
> {code}
> (0 until 6).foldLeft(Seq(1, 2, 3).toDS) { (plan, iteration) =>
> val start = System.currentTimeMillis()
> val result = plan.join(plan, "value").join(plan, "value").join(plan, "value").join(plan, "value")
> result.cache()
> System.out.println(s"Iteration $iteration takes time ${System.currentTimeMillis() - start} ms")
> result.as[Int]
> }
> {code}
> We can see that although all plans are cached, the query planning time still grows exponentially and quickly becomes unbearable.
> {noformat}
> Iteration 0 takes time 9 ms
> Iteration 1 takes time 19 ms
> Iteration 2 takes time 61 ms
> Iteration 3 takes time 219 ms
> Iteration 4 takes time 830 ms
> Iteration 5 takes time 4080 ms
> {noformat}
> Similar scenarios can be found in iterative ML code and significantly affects usability.
> This issue can be fixed by introducing a {{checkpoint()}} method for {{Dataset}} that truncates both the query plan and the lineage of the underlying RDD.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org