You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@spark.apache.org by "ulysses-you (via GitHub)" <gi...@apache.org> on 2023/03/03 03:35:54 UTC
[GitHub] [spark] ulysses-you opened a new pull request, #40262: [SPARK-42651][SQL] Optimize global sort to driver sort
ulysses-you opened a new pull request, #40262:
URL: https://github.com/apache/spark/pull/40262
<!--
Thanks for sending a pull request! Here are some tips for you:
1. If this is your first time, please read our contributor guidelines: https://spark.apache.org/contributing.html
2. Ensure you have added or run the appropriate tests for your PR: https://spark.apache.org/developer-tools.html
3. If the PR is unfinished, add '[WIP]' in your PR title, e.g., '[WIP][SPARK-XXXX] Your PR title ...'.
4. Be sure to keep the PR description updated to reflect all changes.
5. Please write your PR title to summarize what this PR proposes.
6. If possible, provide a concise example to reproduce the issue for a faster review.
7. If you want to add a new configuration, please read the guideline first for naming configurations in
'core/src/main/scala/org/apache/spark/internal/config/ConfigEntry.scala'.
8. If you want to add or modify an error type or message, please read the guideline first in
'core/src/main/resources/error/README.md'.
-->
### What changes were proposed in this pull request?
<!--
Please clarify what changes you are proposing. The purpose of this section is to outline the changes and how this PR fixes the issue.
If possible, please consider writing useful notes for better and faster reviews in your PR. See the examples below.
1. If you refactor some codes with changing classes, showing the class hierarchy will help reviewers.
2. If you fix some SQL features, you can provide some references of other DBMSes.
3. If there is design documentation, please add the link.
4. If there is a discussion in the mailing list, please add the link.
-->
This pr adds a new physical plan `DriverSortExec`. It represents a special case of global `SortExec` which is the root node. Then we can save one shuffle.
Add a new config `spark.sql.execution.driverSortThreshold` to control if it's safe to do sort at driver side. We should make sure the max rows of plan is small enough to avoid oom and preformance regression.
This optimization should work fine since It gets benefits from AQE framework that the max rows of logical plan is accurate and it can be propagated cross most of nodes.
### Why are the changes needed?
<!--
Please clarify why the changes are needed. For instance,
1. If you propose a new API, clarify the use case for a new API.
2. If you fix a bug, you can clarify why it is a bug.
-->
Improve performance
### Does this PR introduce _any_ user-facing change?
<!--
Note that it means *any* user-facing change including all aspects such as the documentation fix.
If yes, please clarify the previous behavior and the change this PR proposes - provide the console output, description and/or an example to show the behavior difference if possible.
If possible, please also clarify if this is a user-facing change compared to the released Spark versions or within the unreleased branches such as master.
If no, write 'No'.
-->
no
### How was this patch tested?
<!--
If tests were added, say they were added here. Please make sure to add some test cases that check the changes thoroughly including negative and positive cases if possible.
If it was tested in a way different from regular unit tests, please clarify how you tested step by step, ideally copy and paste-able, so that other reviewers can test and check, and descendants can verify in the future.
If tests were not added, please describe why they were not added and/or why it was difficult to add.
If benchmark tests were added, please run the benchmarks in GitHub Actions for the consistent environment, and the instructions could accord to: https://spark.apache.org/developer-tools.html#github-workflow-benchmarks.
-->
add test and do some benchmark
Sort 50, 000 rows:
```
OpenJDK 64-Bit Server VM 1.8.0_362-b09 on Linux 5.15.0-1033-azure
Intel(R) Xeon(R) CPU E5-2673 v3 @ 2.40GHz
DriverSort: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------------------------------
Sort 1158 1355 210 0.0 23169.3 1.0X
DriverSort 636 680 55 0.1 12715.2 1.8X
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] mridulm commented on pull request #40262: [SPARK-42651][SQL] Optimize global sort to driver sort
Posted by "mridulm (via GitHub)" <gi...@apache.org>.
mridulm commented on PR #40262:
URL: https://github.com/apache/spark/pull/40262#issuecomment-1467021744
Meta comment: moving this to the driver has potential for destabilizing the application.
In SPARK-36419, we added option to move the final `treeAggregate` to executor given the scale challenges seen.
One option would be to do this in an executor as @zhengruifeng suggested, and that should be much more scalable as a solution.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] github-actions[bot] closed pull request #40262: [SPARK-42651][SQL] Optimize global sort to driver sort
Posted by "github-actions[bot] (via GitHub)" <gi...@apache.org>.
github-actions[bot] closed pull request #40262: [SPARK-42651][SQL] Optimize global sort to driver sort
URL: https://github.com/apache/spark/pull/40262
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] cloud-fan commented on a diff in pull request #40262: [SPARK-42651][SQL] Optimize global sort to driver sort
Posted by "cloud-fan (via GitHub)" <gi...@apache.org>.
cloud-fan commented on code in PR #40262:
URL: https://github.com/apache/spark/pull/40262#discussion_r1134038412
##########
sql/core/src/main/scala/org/apache/spark/sql/execution/DriverSortExec.scala:
##########
@@ -0,0 +1,73 @@
+/*
+ * 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
+
+import java.util.concurrent.TimeUnit.NANOSECONDS
+
+import org.apache.spark.rdd.RDD
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Attribute, NamedExpression, SortOrder, UnsafeProjection}
+import org.apache.spark.sql.catalyst.expressions.codegen.LazilyGeneratedOrdering
+import org.apache.spark.sql.catalyst.plans.physical.{Partitioning, SinglePartition}
+import org.apache.spark.sql.execution.metric.{SQLMetric, SQLMetrics}
+
+/**
+ * A special operator for global [[SortExec]] if it is the root node with/without [[ProjectExec]].
+ */
+case class DriverSortExec(
Review Comment:
have you tried to do a local sort first? I think merging sorted data in the driver side should be faster.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] github-actions[bot] commented on pull request #40262: [SPARK-42651][SQL] Optimize global sort to driver sort
Posted by "github-actions[bot] (via GitHub)" <gi...@apache.org>.
github-actions[bot] commented on PR #40262:
URL: https://github.com/apache/spark/pull/40262#issuecomment-1601849967
We're closing this PR because it hasn't been updated in a while. This isn't a judgement on the merit of the PR in any way. It's just a way of keeping the PR queue manageable.
If you'd like to revive this PR, please reopen it and ask a committer to remove the Stale tag!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] cloud-fan commented on a diff in pull request #40262: [SPARK-42651][SQL] Optimize global sort to driver sort
Posted by "cloud-fan (via GitHub)" <gi...@apache.org>.
cloud-fan commented on code in PR #40262:
URL: https://github.com/apache/spark/pull/40262#discussion_r1133901321
##########
sql/core/src/main/scala/org/apache/spark/sql/execution/DriverSortExec.scala:
##########
@@ -0,0 +1,73 @@
+/*
+ * 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
+
+import java.util.concurrent.TimeUnit.NANOSECONDS
+
+import org.apache.spark.rdd.RDD
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Attribute, NamedExpression, SortOrder, UnsafeProjection}
+import org.apache.spark.sql.catalyst.expressions.codegen.LazilyGeneratedOrdering
+import org.apache.spark.sql.catalyst.plans.physical.{Partitioning, SinglePartition}
+import org.apache.spark.sql.execution.metric.{SQLMetric, SQLMetrics}
+
+/**
+ * A special operator for global [[SortExec]] if it is the root node with/without [[ProjectExec]].
+ */
+case class DriverSortExec(
Review Comment:
`DriverSortExec` must has a local sort below, isn't it?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] ulysses-you commented on a diff in pull request #40262: [SPARK-42651][SQL] Optimize global sort to driver sort
Posted by "ulysses-you (via GitHub)" <gi...@apache.org>.
ulysses-you commented on code in PR #40262:
URL: https://github.com/apache/spark/pull/40262#discussion_r1133945479
##########
sql/core/src/main/scala/org/apache/spark/sql/execution/DriverSortExec.scala:
##########
@@ -0,0 +1,73 @@
+/*
+ * 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
+
+import java.util.concurrent.TimeUnit.NANOSECONDS
+
+import org.apache.spark.rdd.RDD
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Attribute, NamedExpression, SortOrder, UnsafeProjection}
+import org.apache.spark.sql.catalyst.expressions.codegen.LazilyGeneratedOrdering
+import org.apache.spark.sql.catalyst.plans.physical.{Partitioning, SinglePartition}
+import org.apache.spark.sql.execution.metric.{SQLMetric, SQLMetrics}
+
+/**
+ * A special operator for global [[SortExec]] if it is the root node with/without [[ProjectExec]].
+ */
+case class DriverSortExec(
Review Comment:
no, it is the local sort..
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] zhengruifeng commented on pull request #40262: [SPARK-42651][SQL] Optimize global sort to driver sort
Posted by "zhengruifeng (via GitHub)" <gi...@apache.org>.
zhengruifeng commented on PR #40262:
URL: https://github.com/apache/spark/pull/40262#issuecomment-1453092378
just a idea, what about: making `SortExec` requires `SinglePartition`
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] ulysses-you commented on pull request #40262: [SPARK-42651][SQL] Optimize global sort to driver sort
Posted by "ulysses-you (via GitHub)" <gi...@apache.org>.
ulysses-you commented on PR #40262:
URL: https://github.com/apache/spark/pull/40262#issuecomment-1453105982
@zhengruifeng thank you for your thought.
The original idea of driver sort is to avoid one shuffle. Requires SinglePartition seems does not help since it still requires a shuffle.
Besides, finally, the result would go to driver, i.e. `df.sort.collect` (It's the reason I match `ReturnAnswer`), so it should be fine to do at driver. Plus, do sort at driver is not the first code place except driver sort. e.g., the merge function of rdd.takeOrdered. It should be safe if the size of plan is small enough.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] ulysses-you commented on pull request #40262: [SPARK-42651][SQL] Optimize global sort to driver sort
Posted by "ulysses-you (via GitHub)" <gi...@apache.org>.
ulysses-you commented on PR #40262:
URL: https://github.com/apache/spark/pull/40262#issuecomment-1455303198
cc @cloud-fan @viirya thank you
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] cloud-fan commented on a diff in pull request #40262: [SPARK-42651][SQL] Optimize global sort to driver sort
Posted by "cloud-fan (via GitHub)" <gi...@apache.org>.
cloud-fan commented on code in PR #40262:
URL: https://github.com/apache/spark/pull/40262#discussion_r1133447713
##########
sql/core/src/main/scala/org/apache/spark/sql/execution/DriverSortExec.scala:
##########
@@ -0,0 +1,73 @@
+/*
+ * 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
+
+import java.util.concurrent.TimeUnit.NANOSECONDS
+
+import org.apache.spark.rdd.RDD
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Attribute, NamedExpression, SortOrder, UnsafeProjection}
+import org.apache.spark.sql.catalyst.expressions.codegen.LazilyGeneratedOrdering
+import org.apache.spark.sql.catalyst.plans.physical.{Partitioning, SinglePartition}
+import org.apache.spark.sql.execution.metric.{SQLMetric, SQLMetrics}
+
+/**
+ * A special operator for global [[SortExec]] if it is the root node with/without [[ProjectExec]].
+ */
+case class DriverSortExec(
Review Comment:
Can we use `TakeOrderedAndProjectExec` to replace it? They seem to be similar. What if we give a large enough limit to `TakeOrderedAndProjectExec`?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] [spark] ulysses-you commented on a diff in pull request #40262: [SPARK-42651][SQL] Optimize global sort to driver sort
Posted by "ulysses-you (via GitHub)" <gi...@apache.org>.
ulysses-you commented on code in PR #40262:
URL: https://github.com/apache/spark/pull/40262#discussion_r1133461951
##########
sql/core/src/main/scala/org/apache/spark/sql/execution/DriverSortExec.scala:
##########
@@ -0,0 +1,73 @@
+/*
+ * 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
+
+import java.util.concurrent.TimeUnit.NANOSECONDS
+
+import org.apache.spark.rdd.RDD
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Attribute, NamedExpression, SortOrder, UnsafeProjection}
+import org.apache.spark.sql.catalyst.expressions.codegen.LazilyGeneratedOrdering
+import org.apache.spark.sql.catalyst.plans.physical.{Partitioning, SinglePartition}
+import org.apache.spark.sql.execution.metric.{SQLMetric, SQLMetrics}
+
+/**
+ * A special operator for global [[SortExec]] if it is the root node with/without [[ProjectExec]].
+ */
+case class DriverSortExec(
Review Comment:
We have a rule `EliminateLimits` to remove limit if it's large enough, i.e., the actually row count is less than or equal to limit number. So if the `Sort` is the root node that means:
1. the user does not specify the limit
2. the limit has no meaning and has been eliminated
I think if the data size is small enough, `DriverSortExec` is faster than `TakeOrderedAndProjectExec`:
- `DriverSortExec` only sort data once at driver side
- `TakeOrderedAndProjectExec` sort the data at executor side then merge the sorted data at driver side
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org