Merge cockroachdb#50674

50674: opt: modify PushLimitIntoIndexJoin to not directly create limited scans r=DrewKimball a=DrewKimball Previously, the PushLimitIntoIndexJoin rule only matched when the index join input was a constrained scan that could take the limit value as a hard limit. This patch relaxes the match condition to allow any index join with any input. In addition, rather than directly creating a limited scan, the rule now simply pushes the limit on top of the input of the index join. This is useful because index joins can be expensive, and discarding input rows can decrease the cost. It also allows the SplitScansIntoUnionScans rule to fire where previously the index join would have prevented this. Co-authored-by: Drew Kimball <[email protected]>
lunevalex · Jun 26, 2020 · c5133c6 · c5133c6
2 parents 6fec17d + 82f3696
commit c5133c6
Show file tree

Hide file tree

Showing 4 changed files with 110 additions and 110 deletions.
diff --git a/pkg/sql/opt/exec/execbuilder/testdata/select_index_flags b/pkg/sql/opt/exec/execbuilder/testdata/select_index_flags
@@ -142,18 +142,18 @@ index-join    ·             ·
 query TTT
 EXPLAIN SELECT * FROM abcd@{FORCE_INDEX=b,ASC} ORDER BY b DESC LIMIT 5
 ----
-·                     distribution  local
-·                     vectorized    true
-limit                 ·             ·
- │                    count         5
- └── sort             ·             ·
-      │               order         -b
-      └── index-join  ·             ·
-           │          table         abcd@primary
-           │          key columns   a
-           └── scan   ·             ·
-·                     table         abcd@b
-·                     spans         FULL SCAN
+·                    distribution  local
+·                    vectorized    true
+index-join           ·             ·
+ │                   table         abcd@primary
+ │                   key columns   a
+ └── limit           ·             ·
+      │              count         5
+      └── sort       ·             ·
+           │         order         -b
+           └── scan  ·             ·
+·                    table         abcd@b
+·                    spans         FULL SCAN
 
 # Force index cd
 query TTT

diff --git a/pkg/sql/opt/xform/rules/limit.opt b/pkg/sql/opt/xform/rules/limit.opt
@@ -31,23 +31,18 @@
 =>
 (Scan (LimitScanPrivate $scanPrivate $limit $ordering))
 
-# PushLimitIntoIndexJoin pushes a limit through an index join and constructs a
-# new Scan operator that incorporates it. Since index lookup can be expensive,
-# it's always better to discard rows beforehand.
+# PushLimitIntoIndexJoin pushes a limit through an index join. Since index
+# lookup can be expensive, it's always better to discard rows beforehand.
 #
 # TODO(radu): we can similarly push Offset too.
 [PushLimitIntoIndexJoin, Explore]
 (Limit
-    (IndexJoin (Scan $scanPrivate:*) $indexJoinPrivate:*)
-    (Const $limit:* & (IsPositiveInt $limit))
-    $ordering:* &
-        (CanLimitConstrainedScan $scanPrivate $ordering)
+    (IndexJoin $input:* $indexJoinPrivate:*)
+    $limitExpr:(Const $limit:* & (IsPositiveInt $limit))
+    $ordering:*
 )
 =>
-(IndexJoin
-    (Scan (LimitScanPrivate $scanPrivate $limit $ordering))
-    $indexJoinPrivate
-)
+(IndexJoin (Limit $input $limitExpr $ordering) $indexJoinPrivate)
 
 # SplitScanIntoUnionScans splits a scan under a limit into a union of limited
 # scans. Example:

diff --git a/pkg/sql/opt/xform/testdata/external/customer b/pkg/sql/opt/xform/testdata/external/customer
@@ -86,35 +86,35 @@ project
  ├── key: (1)
  ├── fd: (1)-->(2,3,5-8,10)
  ├── ordering: +1
- └── limit
+ └── index-join article
       ├── columns: id:1!null feed:2!null folder:3!null title:5 summary:6 content:7 link:8 read:9!null date:10
-      ├── internal-ordering: +1 opt(9)
       ├── cardinality: [0 - 50]
       ├── key: (1)
       ├── fd: ()-->(9), (1)-->(2,3,5-8,10)
       ├── ordering: +1 opt(9) [actual: +1]
-      ├── index-join article
-      │    ├── columns: id:1!null feed:2!null folder:3!null title:5 summary:6 content:7 link:8 read:9!null date:10
-      │    ├── key: (1)
-      │    ├── fd: ()-->(9), (1)-->(2,3,5-8,10)
-      │    ├── ordering: +1 opt(9) [actual: +1]
-      │    ├── limit hint: 50.00
-      │    └── select
-      │         ├── columns: id:1!null feed:2!null folder:3!null read:9!null
-      │         ├── key: (1)
-      │         ├── fd: ()-->(9), (1)-->(2,3)
-      │         ├── ordering: +1 opt(9) [actual: +1]
-      │         ├── limit hint: 50.00
-      │         ├── scan article@article_idx_read_key
-      │         │    ├── columns: id:1!null feed:2!null folder:3!null read:9
-      │         │    ├── constraint: /1: [/1 - ]
-      │         │    ├── key: (1)
-      │         │    ├── fd: (1)-->(2,3,9)
-      │         │    ├── ordering: +1 opt(9) [actual: +1]
-      │         │    └── limit hint: 101.01
-      │         └── filters
-      │              └── NOT read:9 [outer=(9), constraints=(/9: [/false - /false]; tight), fd=()-->(9)]
-      └── 50
+      └── limit
+           ├── columns: id:1!null feed:2!null folder:3!null read:9!null
+           ├── internal-ordering: +1 opt(9)
+           ├── cardinality: [0 - 50]
+           ├── key: (1)
+           ├── fd: ()-->(9), (1)-->(2,3)
+           ├── ordering: +1 opt(9) [actual: +1]
+           ├── select
+           │    ├── columns: id:1!null feed:2!null folder:3!null read:9!null
+           │    ├── key: (1)
+           │    ├── fd: ()-->(9), (1)-->(2,3)
+           │    ├── ordering: +1 opt(9) [actual: +1]
+           │    ├── limit hint: 50.00
+           │    ├── scan article@article_idx_read_key
+           │    │    ├── columns: id:1!null feed:2!null folder:3!null read:9
+           │    │    ├── constraint: /1: [/1 - ]
+           │    │    ├── key: (1)
+           │    │    ├── fd: (1)-->(2,3,9)
+           │    │    ├── ordering: +1 opt(9) [actual: +1]
+           │    │    └── limit hint: 101.01
+           │    └── filters
+           │         └── NOT read:9 [outer=(9), constraints=(/9: [/false - /false]; tight), fd=()-->(9)]
+           └── 50
 
 # Check that forcing the index works as well.
 opt
@@ -128,36 +128,36 @@ project
  ├── key: (1)
  ├── fd: (1)-->(2,3,5-8,10)
  ├── ordering: +1
- └── limit
+ └── index-join article
       ├── columns: id:1!null feed:2!null folder:3!null title:5 summary:6 content:7 link:8 read:9!null date:10
-      ├── internal-ordering: +1 opt(9)
       ├── cardinality: [0 - 50]
       ├── key: (1)
       ├── fd: ()-->(9), (1)-->(2,3,5-8,10)
       ├── ordering: +1 opt(9) [actual: +1]
-      ├── index-join article
-      │    ├── columns: id:1!null feed:2!null folder:3!null title:5 summary:6 content:7 link:8 read:9!null date:10
-      │    ├── key: (1)
-      │    ├── fd: ()-->(9), (1)-->(2,3,5-8,10)
-      │    ├── ordering: +1 opt(9) [actual: +1]
-      │    ├── limit hint: 50.00
-      │    └── select
-      │         ├── columns: id:1!null feed:2!null folder:3!null read:9!null
-      │         ├── key: (1)
-      │         ├── fd: ()-->(9), (1)-->(2,3)
-      │         ├── ordering: +1 opt(9) [actual: +1]
-      │         ├── limit hint: 50.00
-      │         ├── scan article@article_idx_read_key
-      │         │    ├── columns: id:1!null feed:2!null folder:3!null read:9
-      │         │    ├── constraint: /1: [/1 - ]
-      │         │    ├── flags: force-index=article_idx_read_key
-      │         │    ├── key: (1)
-      │         │    ├── fd: (1)-->(2,3,9)
-      │         │    ├── ordering: +1 opt(9) [actual: +1]
-      │         │    └── limit hint: 101.01
-      │         └── filters
-      │              └── NOT read:9 [outer=(9), constraints=(/9: [/false - /false]; tight), fd=()-->(9)]
-      └── 50
+      └── limit
+           ├── columns: id:1!null feed:2!null folder:3!null read:9!null
+           ├── internal-ordering: +1 opt(9)
+           ├── cardinality: [0 - 50]
+           ├── key: (1)
+           ├── fd: ()-->(9), (1)-->(2,3)
+           ├── ordering: +1 opt(9) [actual: +1]
+           ├── select
+           │    ├── columns: id:1!null feed:2!null folder:3!null read:9!null
+           │    ├── key: (1)
+           │    ├── fd: ()-->(9), (1)-->(2,3)
+           │    ├── ordering: +1 opt(9) [actual: +1]
+           │    ├── limit hint: 50.00
+           │    ├── scan article@article_idx_read_key
+           │    │    ├── columns: id:1!null feed:2!null folder:3!null read:9
+           │    │    ├── constraint: /1: [/1 - ]
+           │    │    ├── flags: force-index=article_idx_read_key
+           │    │    ├── key: (1)
+           │    │    ├── fd: (1)-->(2,3,9)
+           │    │    ├── ordering: +1 opt(9) [actual: +1]
+           │    │    └── limit hint: 101.01
+           │    └── filters
+           │         └── NOT read:9 [outer=(9), constraints=(/9: [/false - /false]; tight), fd=()-->(9)]
+           └── 50
 
 # Use only columns covered by the index.
 opt

diff --git a/pkg/sql/opt/xform/testdata/rules/limit b/pkg/sql/opt/xform/testdata/rules/limit
@@ -229,36 +229,39 @@ index-join kuv
       ├── fd: (1)-->(2)
       └── ordering: +2
 
-# Verify we don't push the limit if the ordering depends on a column not in the
-# input index.
 opt
 SELECT * FROM kuv WHERE u > 1 AND u < 10 ORDER BY u, v LIMIT 5
 ----
-limit
+sort
  ├── columns: k:1!null u:2!null v:3
- ├── internal-ordering: +2,+3
  ├── cardinality: [0 - 5]
  ├── key: (1)
  ├── fd: (1)-->(2,3)
  ├── ordering: +2,+3
- ├── sort (segmented)
- │    ├── columns: k:1!null u:2!null v:3
- │    ├── key: (1)
- │    ├── fd: (1)-->(2,3)
- │    ├── ordering: +2,+3
- │    ├── limit hint: 5.00
- │    └── index-join kuv
- │         ├── columns: k:1!null u:2!null v:3
- │         ├── key: (1)
- │         ├── fd: (1)-->(2,3)
- │         ├── ordering: +2
- │         └── scan kuv@secondary
- │              ├── columns: k:1!null u:2!null
- │              ├── constraint: /2/1: [/2 - /9]
- │              ├── key: (1)
- │              ├── fd: (1)-->(2)
- │              └── ordering: +2
- └── 5
+ └── index-join kuv
+      ├── columns: k:1!null u:2!null v:3
+      ├── cardinality: [0 - 5]
+      ├── key: (1)
+      ├── fd: (1)-->(2,3)
+      └── limit
+           ├── columns: k:1!null u:2!null
+           ├── internal-ordering: +2,+3
+           ├── cardinality: [0 - 5]
+           ├── key: (1)
+           ├── fd: (1)-->(2)
+           ├── sort (segmented)
+           │    ├── columns: k:1!null u:2!null
+           │    ├── key: (1)
+           │    ├── fd: (1)-->(2)
+           │    ├── ordering: +2,+3
+           │    ├── limit hint: 5.00
+           │    └── scan kuv@secondary
+           │         ├── columns: k:1!null u:2!null
+           │         ├── constraint: /2/1: [/2 - /9]
+           │         ├── key: (1)
+           │         ├── fd: (1)-->(2)
+           │         └── ordering: +2
+           └── 5
 
 exec-ddl
 CREATE TABLE abcd (
@@ -719,6 +722,24 @@ limit
  │              └── limit: 10
  └── 10
 
+# Case with index join.
+opt expect=SplitScanIntoUnionScans format=hide-all
+SELECT *
+FROM index_tab WHERE region = 'US_WEST' OR region = 'US_EAST'
+ORDER BY data1 LIMIT 10
+----
+index-join index_tab
+ └── limit
+      ├── sort
+      │    └── union
+      │         ├── scan index_tab@c
+      │         │    ├── constraint: /10/13/14/8: [/'US_EAST' - /'US_EAST']
+      │         │    └── limit: 10
+      │         └── scan index_tab@c
+      │              ├── constraint: /17/20/21/15: [/'US_WEST' - /'US_WEST']
+      │              └── limit: 10
+      └── 10
+
 # No-op case because the multi-key span isn't countable.
 opt expect-not=SplitScanIntoUnionScans format=hide-all
 SELECT max(data1) FROM index_tab WHERE region > 'US_EAST' AND region < 'US_WEST'
@@ -793,19 +814,3 @@ scalar-group-by
  └── aggregations
       └── max
            └── data1
-
-# No-op case because the limit input is an index-join. TODO(drewk): modify
-# the rule to push limits through index joins.
-opt expect-not=SplitScanIntoUnionScans format=hide-all
-SELECT *
-FROM index_tab WHERE region = 'US_WEST' OR region = 'US_EAST'
-ORDER BY data1 LIMIT 10
-----
-limit
- ├── sort
- │    └── index-join index_tab
- │         └── scan index_tab@c
- │              └── constraint: /3/6/7/1
- │                   ├── [/'US_EAST' - /'US_EAST']
- │                   └── [/'US_WEST' - /'US_WEST']
- └── 10