mirror of
https://github.com/langchain-ai/datafusion.git
synced 2026-07-21 03:05:28 -04:00
Gene Bordegaray
552dbe4f19
## Which issue does this PR close? <!-- We generally require a GitHub issue to be filed for all bug fixes and enhancements and this helps us generate change logs for our releases. You can link an issue to this PR using the GitHub syntax. For example `Closes #123` indicates that this PR will close issue #123. --> - Closes #18341. - Closes https://github.com/apache/datafusion/issues/9370 ## Rationale for this change <!-- Why are you proposing this change? If this is already explained clearly in the issue then this section is not needed. Explaining clearly why changes are proposed helps reviewers understand your changes and offer better suggestions for fixes. --> Cases where two RepartitionExec operators appear consecutively in the plan. This is unneeded overhead that eliminating provides speed ups. Full Report: [The Physical Optimizer and Fixing Consecutive Repartitions In the Enforce Distribution Rule.pdf](https://github.com/user-attachments/files/23420831/The.Physical.Optimizer.and.Fixing.Consecutive.Repartitions.In.the.Enforce.Distribution.Rule.pdf) Issue Report: [Fixing Consecutive Repartitions In the Enforce Distribution Rule.pdf](https://github.com/user-attachments/files/23420880/Fixing.Consecutive.Repartitions.In.the.Enforce.Distribution.Rule.pdf) ## What changes are included in this PR? <!-- There is no need to duplicate the description in the issue here but it is sometimes worth providing a summary of the individual changes in this PR. --> Change to repartition adding logic in `enforce_distribution.rs` A ton of test and bench updates to mirror new behavior ## Are these changes tested? <!-- We typically require tests for all PRs in order to: 1. Prevent the code from being accidentally broken by subsequent changes 2. Serve as another way to document the expected behavior of the code If tests are not included in your PR, please explain why (for example, are they covered by existing tests)? --> Yes benchmarked and tested, check report for benchmarks ## Are there any user-facing changes? <!-- If there are user-facing changes then we may require documentation to be updated before approving the PR. --> <!-- If there are any breaking changes to public APIs, please add the `api change` label. --> --------- Co-authored-by: Andrew Lamb <andrew@nerdnetworks.org>
17 KiB
17 KiB
EXPLAIN
The EXPLAIN command shows the logical and physical execution plan for the specified SQL statement.
Syntax
EXPLAIN [ANALYZE] [VERBOSE] [FORMAT format] statement
EXPLAIN
Shows the execution plan of a statement.
If you need more detailed output, use EXPLAIN VERBOSE.
Note that EXPLAIN VERBOSE only supports the indent format.
The optional [FORMAT format] clause controls how the plan is displayed as
explained below. If this clause is not specified, the plan is displayed using
the format from the configuration value datafusion.explain.format.
tree format (default)
The tree format is modeled after DuckDB plans and is designed to be easier
to see the high level structure of the plan
> EXPLAIN FORMAT TREE SELECT SUM(x) FROM t GROUP BY b;
+---------------+-------------------------------+
| plan_type | plan |
+---------------+-------------------------------+
| physical_plan | ┌───────────────────────────┐ |
| | │ ProjectionExec │ |
| | │ -------------------- │ |
| | │ sum(t.x): sum(t.x)@1 │ |
| | └─────────────┬─────────────┘ |
| | ┌─────────────┴─────────────┐ |
| | │ AggregateExec │ |
| | │ -------------------- │ |
| | │ aggr: sum(t.x) │ |
| | │ group_by: b@0 as b │ |
| | │ │ |
| | │ mode: │ |
| | │ FinalPartitioned │ |
| | └─────────────┬─────────────┘ |
| | ┌─────────────┴─────────────┐ |
| | │ CoalesceBatchesExec │ |
| | └─────────────┬─────────────┘ |
| | ┌─────────────┴─────────────┐ |
| | │ RepartitionExec │ |
| | │ -------------------- │ |
| | │ input_partition_count: │ |
| | │ 1 │ |
| | │ │ |
| | │ partitioning_scheme: │ |
| | │ Hash([b@0], 16) │ |
| | └─────────────┬─────────────┘ |
| | ┌─────────────┴─────────────┐ |
| | │ AggregateExec │ |
| | │ -------------------- │ |
| | │ aggr: sum(t.x) │ |
| | │ group_by: b@1 as b │ |
| | │ mode: Partial │ |
| | └─────────────┬─────────────┘ |
| | ┌─────────────┴─────────────┐ |
| | │ DataSourceExec │ |
| | │ -------------------- │ |
| | │ bytes: 224 │ |
| | │ format: memory │ |
| | │ rows: 1 │ |
| | └───────────────────────────┘ |
| | |
+---------------+-------------------------------+
1 row(s) fetched.
Elapsed 0.016 seconds.
indent format
The indent format shows both the logical and physical plan, with one line for
each operator in the plan. Child plans are indented to show the hierarchy.
See Reading Explain Plans for more information on how to interpret these plans.
> CREATE TABLE t(x int, b int) AS VALUES (1, 2), (2, 3);
0 row(s) fetched.
Elapsed 0.004 seconds.
> EXPLAIN FORMAT INDENT SELECT SUM(x) FROM t GROUP BY b;
+---------------+-------------------------------------------------------------------------------+
| plan_type | plan |
+---------------+-------------------------------------------------------------------------------+
| logical_plan | Projection: sum(t.x) |
| | Aggregate: groupBy=[[t.b]], aggr=[[sum(CAST(t.x AS Int64))]] |
| | TableScan: t projection=[x, b] |
| physical_plan | ProjectionExec: expr=[sum(t.x)@1 as sum(t.x)] |
| | AggregateExec: mode=FinalPartitioned, gby=[b@0 as b], aggr=[sum(t.x)] |
| | CoalesceBatchesExec: target_batch_size=8192 |
| | RepartitionExec: partitioning=Hash([b@0], 16), input_partitions=1 |
| | AggregateExec: mode=Partial, gby=[b@1 as b], aggr=[sum(t.x)] |
| | DataSourceExec: partitions=1, partition_sizes=[1] |
| | |
+---------------+-------------------------------------------------------------------------------+
2 row(s) fetched.
Elapsed 0.004 seconds.
pgjson format
The pgjson format is modeled after Postgres JSON format.
You can use this format to visualize the plan in existing plan visualization tools, such as dalibo
> EXPLAIN FORMAT PGJSON SELECT SUM(x) FROM t GROUP BY b;
+--------------+----------------------------------------------------+
| plan_type | plan |
+--------------+----------------------------------------------------+
| logical_plan | [ |
| | { |
| | "Plan": { |
| | "Expressions": [ |
| | "sum(t.x)" |
| | ], |
| | "Node Type": "Projection", |
| | "Output": [ |
| | "sum(t.x)" |
| | ], |
| | "Plans": [ |
| | { |
| | "Aggregates": "sum(CAST(t.x AS Int64))", |
| | "Group By": "t.b", |
| | "Node Type": "Aggregate", |
| | "Output": [ |
| | "b", |
| | "sum(t.x)" |
| | ], |
| | "Plans": [ |
| | { |
| | "Node Type": "TableScan", |
| | "Output": [ |
| | "x", |
| | "b" |
| | ], |
| | "Plans": [], |
| | "Relation Name": "t" |
| | } |
| | ] |
| | } |
| | ] |
| | } |
| | } |
| | ] |
+--------------+----------------------------------------------------+
1 row(s) fetched.
Elapsed 0.008 seconds.
graphviz format
The graphviz format uses the DOT language that can be used with Graphviz to
generate a visual representation of the plan.
> EXPLAIN FORMAT GRAPHVIZ SELECT SUM(x) FROM t GROUP BY b;
+--------------+------------------------------------------------------------------------------------------------------------------------------+
| plan_type | plan |
+--------------+------------------------------------------------------------------------------------------------------------------------------+
| logical_plan | |
| | // Begin DataFusion GraphViz Plan, |
| | // display it online here: https://dreampuf.github.io/GraphvizOnline |
| | |
| | digraph { |
| | subgraph cluster_1 |
| | { |
| | graph[label="LogicalPlan"] |
| | 2[shape=box label="Projection: sum(t.x)"] |
| | 3[shape=box label="Aggregate: groupBy=[[t.b]], aggr=[[sum(CAST(t.x AS Int64))]]"] |
| | 2 -> 3 [arrowhead=none, arrowtail=normal, dir=back] |
| | 4[shape=box label="TableScan: t projection=[x, b]"] |
| | 3 -> 4 [arrowhead=none, arrowtail=normal, dir=back] |
| | } |
| | subgraph cluster_5 |
| | { |
| | graph[label="Detailed LogicalPlan"] |
| | 6[shape=box label="Projection: sum(t.x)\nSchema: [sum(t.x):Int64;N]"] |
| | 7[shape=box label="Aggregate: groupBy=[[t.b]], aggr=[[sum(CAST(t.x AS Int64))]]\nSchema: [b:Int32;N, sum(t.x):Int64;N]"] |
| | 6 -> 7 [arrowhead=none, arrowtail=normal, dir=back] |
| | 8[shape=box label="TableScan: t projection=[x, b]\nSchema: [x:Int32;N, b:Int32;N]"] |
| | 7 -> 8 [arrowhead=none, arrowtail=normal, dir=back] |
| | } |
| | } |
| | // End DataFusion GraphViz Plan |
| | |
+--------------+------------------------------------------------------------------------------------------------------------------------------+
1 row(s) fetched.
Elapsed 0.010 seconds.
EXPLAIN ANALYZE
Shows the execution plan and metrics of a statement. Note that EXPLAIN ANALYZE
only supports the indent format.
EXPLAIN ANALYZE SELECT SUM(x) FROM table GROUP BY b;
+-------------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------+
| plan_type | plan |
+-------------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------+
| Plan with Metrics | CoalescePartitionsExec, metrics=[] |
| | ProjectionExec: expr=[SUM(table.x)@1 as SUM(x)], metrics=[] |
| | HashAggregateExec: mode=FinalPartitioned, gby=[b@0 as b], aggr=[SUM(x)], metrics=[outputRows=2] |
| | CoalesceBatchesExec: target_batch_size=4096, metrics=[] |
| | RepartitionExec: partitioning=Hash([Column { name: "b", index: 0 }], 16), metrics=[sendTime=839560, fetchTime=122528525, repartitionTime=5327877] |
| | HashAggregateExec: mode=Partial, gby=[b@1 as b], aggr=[SUM(x)], metrics=[outputRows=2] |
| | RepartitionExec: partitioning=RoundRobinBatch(16), metrics=[fetchTime=5660489, repartitionTime=0, sendTime=8012] |
| | DataSourceExec: file_groups={1 group: [[/tmp/table.csv]]}, has_header=false, metrics=[] |
+-------------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------+
By default EXPLAIN ANALYZE shows the aggregated metrics from all partitions for each operator. If you need to display per-partition metrics, use EXPLAIN ANALYZE VERBOSE.
You can also set datafusion.explain.analyze_level from the configuration value to control the detail level for the metrics displayed.