From 64bad8e2f7da4569df53709ad2f777c4e27835db Mon Sep 17 00:00:00 2001
From: Marcus Gartner <marcus@cockroachlabs.com>
Date: Tue, 28 Jan 2025 16:16:49 -0500
Subject: [PATCH] opt: add `optimizer_prefer_bounded_cardinality` session
setting
Release note (sql change): The `optimizer_prefer_bounded_cardinality`
session setting has been added which instructs the optimizer to prefer
query plans where every expression has a guaranteed upper-bound on the
number of rows it will process. This may help the optimizer produce
better query plans in some cases. This setting is disabled by default.
---
pkg/sql/exec_util.go | 4 +
.../testdata/logic_test/information_schema | 1 +
.../logictest/testdata/logic_test/pg_catalog | 3 +
.../logictest/testdata/logic_test/show_source | 1 +
pkg/sql/opt/memo/cost.go | 18 +-
pkg/sql/opt/memo/cost_test.go | 4 +
pkg/sql/opt/memo/expr_format.go | 3 +
pkg/sql/opt/memo/memo.go | 3 +
pkg/sql/opt/memo/memo_test.go | 5 +
pkg/sql/opt/xform/coster.go | 5 +
.../xform/testdata/coster/outside-histogram | 484 ++++++++++++++++++
pkg/sql/prepared_stmt.go | 5 +
.../local_only_session_data.proto | 4 +
pkg/sql/vars.go | 18 +
14 files changed, 555 insertions(+), 3 deletions(-)
create mode 100644 pkg/sql/opt/xform/testdata/coster/outside-histogram
diff --git a/pkg/sql/exec_util.go b/pkg/sql/exec_util.go
index a8028c2282b1..8c057b18dcb8 100644
--- a/pkg/sql/exec_util.go
+++ b/pkg/sql/exec_util.go
@@ -4007,6 +4007,10 @@ func (m *sessionDataMutator) SetCatalogDigestStalenessCheckEnabled(b bool) {
m.data.CatalogDigestStalenessCheckEnabled = b
}
+func (m *sessionDataMutator) SetOptimizerPreferBoundedCardinality(b bool) {
+ m.data.OptimizerPreferBoundedCardinality = b
+}
+
// Utility functions related to scrubbing sensitive information on SQL Stats.
// quantizeCounts ensures that the Count field in the
diff --git a/pkg/sql/logictest/testdata/logic_test/information_schema b/pkg/sql/logictest/testdata/logic_test/information_schema
index c6edb6e13b2c..c1bd12635769 100644
--- a/pkg/sql/logictest/testdata/logic_test/information_schema
+++ b/pkg/sql/logictest/testdata/logic_test/information_schema
@@ -4007,6 +4007,7 @@ optimizer on
optimizer_always_use_histograms on
optimizer_hoist_uncorrelated_equality_subqueries on
optimizer_merge_joins_enabled on
+optimizer_prefer_bounded_cardinality off
optimizer_prove_implication_with_virtual_computed_columns on
optimizer_push_limit_into_project_filtered_scan on
optimizer_push_offset_into_index_join on
diff --git a/pkg/sql/logictest/testdata/logic_test/pg_catalog b/pkg/sql/logictest/testdata/logic_test/pg_catalog
index 8499f75e0678..9c8346704a77 100644
--- a/pkg/sql/logictest/testdata/logic_test/pg_catalog
+++ b/pkg/sql/logictest/testdata/logic_test/pg_catalog
@@ -3007,6 +3007,7 @@ opt_split_scan_limit 2048 N
optimizer_always_use_histograms on NULL NULL NULL string
optimizer_hoist_uncorrelated_equality_subqueries on NULL NULL NULL string
optimizer_merge_joins_enabled on NULL NULL NULL string
+optimizer_prefer_bounded_cardinality off NULL NULL NULL string
optimizer_prove_implication_with_virtual_computed_columns on NULL NULL NULL string
optimizer_push_limit_into_project_filtered_scan on NULL NULL NULL string
optimizer_push_offset_into_index_join on NULL NULL NULL string
@@ -3212,6 +3213,7 @@ opt_split_scan_limit 2048 N
optimizer_always_use_histograms on NULL user NULL on on
optimizer_hoist_uncorrelated_equality_subqueries on NULL user NULL on on
optimizer_merge_joins_enabled on NULL user NULL on on
+optimizer_prefer_bounded_cardinality off NULL user NULL off off
optimizer_prove_implication_with_virtual_computed_columns on NULL user NULL on on
optimizer_push_limit_into_project_filtered_scan on NULL user NULL on on
optimizer_push_offset_into_index_join on NULL user NULL on on
@@ -3416,6 +3418,7 @@ optimizer NULL NULL NULL
optimizer_always_use_histograms NULL NULL NULL NULL NULL
optimizer_hoist_uncorrelated_equality_subqueries NULL NULL NULL NULL NULL
optimizer_merge_joins_enabled NULL NULL NULL NULL NULL
+optimizer_prefer_bounded_cardinality NULL NULL NULL NULL NULL
optimizer_prove_implication_with_virtual_computed_columns NULL NULL NULL NULL NULL
optimizer_push_limit_into_project_filtered_scan NULL NULL NULL NULL NULL
optimizer_push_offset_into_index_join NULL NULL NULL NULL NULL
diff --git a/pkg/sql/logictest/testdata/logic_test/show_source b/pkg/sql/logictest/testdata/logic_test/show_source
index b884b2f2a6f4..8ae571dd5dfe 100644
--- a/pkg/sql/logictest/testdata/logic_test/show_source
+++ b/pkg/sql/logictest/testdata/logic_test/show_source
@@ -138,6 +138,7 @@ opt_split_scan_limit 2048
optimizer_always_use_histograms on
optimizer_hoist_uncorrelated_equality_subqueries on
optimizer_merge_joins_enabled on
+optimizer_prefer_bounded_cardinality off
optimizer_prove_implication_with_virtual_computed_columns on
optimizer_push_limit_into_project_filtered_scan on
optimizer_push_offset_into_index_join on
diff --git a/pkg/sql/opt/memo/cost.go b/pkg/sql/opt/memo/cost.go
index 6dfd70265abc..8dbcda97903a 100644
--- a/pkg/sql/opt/memo/cost.go
+++ b/pkg/sql/opt/memo/cost.go
@@ -19,8 +19,12 @@ type Cost struct {
// group members during testing, by setting their cost so high that any other
// member will have a lower cost.
var MaxCost = Cost{
- C: math.Inf(+1),
- Flags: CostFlags{FullScanPenalty: true, HugeCostPenalty: true},
+ C: math.Inf(+1),
+ Flags: CostFlags{
+ FullScanPenalty: true,
+ HugeCostPenalty: true,
+ UnboundedCardinality: true,
+ },
}
// Less returns true if this cost is lower than the given cost.
@@ -57,6 +61,10 @@ type CostFlags struct {
// used when the optimizer is forced to use a particular plan, and will error
// if it cannot be used.
HugeCostPenalty bool
+ // UnboundedCardinality is true if the operator or any of its descendants
+ // have no guaranteed upperbound on the number of rows that they can
+ // produce. See props.AnyCardinality.
+ UnboundedCardinality bool
}
// Less returns true if these flags indicate a lower penalty than the other
@@ -71,6 +79,9 @@ func (c CostFlags) Less(other CostFlags) bool {
if c.FullScanPenalty != other.FullScanPenalty {
return !c.FullScanPenalty
}
+ if c.UnboundedCardinality != other.UnboundedCardinality {
+ return !c.UnboundedCardinality
+ }
return false
}
@@ -78,9 +89,10 @@ func (c CostFlags) Less(other CostFlags) bool {
func (c *CostFlags) Add(other CostFlags) {
c.FullScanPenalty = c.FullScanPenalty || other.FullScanPenalty
c.HugeCostPenalty = c.HugeCostPenalty || other.HugeCostPenalty
+ c.UnboundedCardinality = c.UnboundedCardinality || other.UnboundedCardinality
}
// Empty returns true if these flags are empty.
func (c CostFlags) Empty() bool {
- return !c.FullScanPenalty && !c.HugeCostPenalty
+ return !c.FullScanPenalty && !c.HugeCostPenalty && !c.UnboundedCardinality
}
diff --git a/pkg/sql/opt/memo/cost_test.go b/pkg/sql/opt/memo/cost_test.go
index ed8235731ad2..c44e3247c7ad 100644
--- a/pkg/sql/opt/memo/cost_test.go
+++ b/pkg/sql/opt/memo/cost_test.go
@@ -35,6 +35,8 @@ func TestCostLess(t *testing.T) {
{memo.MaxCost, memo.MaxCost, false},
{memo.MaxCost, memo.Cost{C: 1.0, Flags: memo.CostFlags{FullScanPenalty: true}}, false},
{memo.Cost{C: 1.0, Flags: memo.CostFlags{HugeCostPenalty: true}}, memo.MaxCost, true},
+ {memo.Cost{C: 2.0, Flags: memo.CostFlags{}}, memo.Cost{C: 1.0, Flags: memo.CostFlags{UnboundedCardinality: true}}, true},
+ {memo.Cost{C: 1.0, Flags: memo.CostFlags{UnboundedCardinality: true}}, memo.Cost{C: 2.0, Flags: memo.CostFlags{}}, false},
}
for _, tc := range testCases {
if tc.left.Less(tc.right) != tc.expected {
@@ -72,6 +74,8 @@ func TestCostFlagsLess(t *testing.T) {
{memo.CostFlags{FullScanPenalty: true, HugeCostPenalty: true}, memo.CostFlags{FullScanPenalty: true, HugeCostPenalty: true}, false},
{memo.CostFlags{FullScanPenalty: false}, memo.CostFlags{FullScanPenalty: true}, true},
{memo.CostFlags{HugeCostPenalty: false}, memo.CostFlags{HugeCostPenalty: true}, true},
+ {memo.CostFlags{UnboundedCardinality: false}, memo.CostFlags{UnboundedCardinality: true}, true},
+ {memo.CostFlags{UnboundedCardinality: true}, memo.CostFlags{UnboundedCardinality: false}, false},
}
for _, tc := range testCases {
if tc.left.Less(tc.right) != tc.expected {
diff --git a/pkg/sql/opt/memo/expr_format.go b/pkg/sql/opt/memo/expr_format.go
index 2029e911a3a0..adc2af70d31d 100644
--- a/pkg/sql/opt/memo/expr_format.go
+++ b/pkg/sql/opt/memo/expr_format.go
@@ -913,6 +913,9 @@ func (f *ExprFmtCtx) formatRelational(e RelExpr, tp treeprinter.Node) {
if cost.Flags.HugeCostPenalty {
b.WriteString(" huge-cost-penalty")
}
+ if cost.Flags.UnboundedCardinality {
+ b.WriteString(" unbounded-cardinality")
+ }
tp.Child(b.String())
}
}
diff --git a/pkg/sql/opt/memo/memo.go b/pkg/sql/opt/memo/memo.go
index fa24b707c9e1..5bf19798d081 100644
--- a/pkg/sql/opt/memo/memo.go
+++ b/pkg/sql/opt/memo/memo.go
@@ -200,6 +200,7 @@ type Memo struct {
pushLimitIntoProjectFilteredScan bool
unsafeAllowTriggersModifyingCascades bool
legacyVarcharTyping bool
+ preferBoundedCardinality bool
internal bool
// txnIsoLevel is the isolation level under which the plan was created. This
@@ -293,6 +294,7 @@ func (m *Memo) Init(ctx context.Context, evalCtx *eval.Context) {
pushLimitIntoProjectFilteredScan: evalCtx.SessionData().OptimizerPushLimitIntoProjectFilteredScan,
unsafeAllowTriggersModifyingCascades: evalCtx.SessionData().UnsafeAllowTriggersModifyingCascades,
legacyVarcharTyping: evalCtx.SessionData().LegacyVarcharTyping,
+ preferBoundedCardinality: evalCtx.SessionData().OptimizerPreferBoundedCardinality,
internal: evalCtx.SessionData().Internal,
txnIsoLevel: evalCtx.TxnIsoLevel,
}
@@ -463,6 +465,7 @@ func (m *Memo) IsStale(
m.pushLimitIntoProjectFilteredScan != evalCtx.SessionData().OptimizerPushLimitIntoProjectFilteredScan ||
m.unsafeAllowTriggersModifyingCascades != evalCtx.SessionData().UnsafeAllowTriggersModifyingCascades ||
m.legacyVarcharTyping != evalCtx.SessionData().LegacyVarcharTyping ||
+ m.preferBoundedCardinality != evalCtx.SessionData().OptimizerPreferBoundedCardinality ||
m.internal != evalCtx.SessionData().Internal ||
m.txnIsoLevel != evalCtx.TxnIsoLevel {
return true, nil
diff --git a/pkg/sql/opt/memo/memo_test.go b/pkg/sql/opt/memo/memo_test.go
index 34f654726a15..65fc371965f7 100644
--- a/pkg/sql/opt/memo/memo_test.go
+++ b/pkg/sql/opt/memo/memo_test.go
@@ -538,6 +538,11 @@ func TestMemoIsStale(t *testing.T) {
evalCtx.SessionData().LegacyVarcharTyping = false
notStale()
+ evalCtx.SessionData().OptimizerPreferBoundedCardinality = true
+ stale()
+ evalCtx.SessionData().OptimizerPreferBoundedCardinality = false
+ notStale()
+
evalCtx.SessionData().Internal = true
stale()
evalCtx.SessionData().Internal = false
diff --git a/pkg/sql/opt/xform/coster.go b/pkg/sql/opt/xform/coster.go
index 07391cf4724e..f5a59789fbf2 100644
--- a/pkg/sql/opt/xform/coster.go
+++ b/pkg/sql/opt/xform/coster.go
@@ -625,8 +625,13 @@ func (c *coster) ComputeCost(candidate memo.RelExpr, required *physical.Required
// Add a one-time cost for any operator with unbounded cardinality. This
// ensures we prefer plans that push limits as far down the tree as possible,
// all else being equal.
+ //
+ // Also add a cost flag for unbounded cardinality.
if candidate.Relational().Cardinality.IsUnbounded() {
cost.C += cpuCostFactor
+ if c.evalCtx.SessionData().OptimizerPreferBoundedCardinality {
+ cost.Flags.UnboundedCardinality = true
+ }
}
if !cost.Less(memo.MaxCost) {
diff --git a/pkg/sql/opt/xform/testdata/coster/outside-histogram b/pkg/sql/opt/xform/testdata/coster/outside-histogram
new file mode 100644
index 000000000000..0acd1fe544f4
--- /dev/null
+++ b/pkg/sql/opt/xform/testdata/coster/outside-histogram
@@ -0,0 +1,484 @@
+# This file contains tests for queries that filter columns beyond the max/min
+# values in their histograms. Each query is tested multiple times with different
+# settings to show the effect those settings have on the plan.
+#
+
+exec-ddl
+CREATE TABLE t (
+ k INT PRIMARY KEY,
+ i INT,
+ s STRING,
+ INDEX (i),
+ INDEX (s)
+)
+----
+
+exec-ddl
+ALTER TABLE t INJECT STATISTICS '[
+ {
+ "columns": ["k"],
+ "created_at": "2018-01-01 1:00:00.00000+00:00",
+ "row_count": 1000,
+ "distinct_count": 1000,
+ "null_count": 0,
+ "avg_size": 2,
+ "histo_col_type": "int",
+ "histo_buckets": [
+ {"num_eq": 0, "num_range": 0, "distinct_range": 0, "upper_bound": "0"},
+ {"num_eq": 1, "num_range": 99, "distinct_range": 99, "upper_bound": "100"},
+ {"num_eq": 1, "num_range": 199, "distinct_range": 199, "upper_bound": "200"},
+ {"num_eq": 1, "num_range": 299, "distinct_range": 299, "upper_bound": "300"},
+ {"num_eq": 1, "num_range": 399, "distinct_range": 399, "upper_bound": "400"}
+ ]
+ },
+ {
+ "columns": ["i"],
+ "created_at": "2018-01-01 1:00:00.00000+00:00",
+ "row_count": 1000,
+ "distinct_count": 44,
+ "null_count": 0,
+ "avg_size": 2,
+ "histo_col_type": "int",
+ "histo_buckets": [
+ {"num_eq": 0, "num_range": 0, "distinct_range": 0, "upper_bound": "0"},
+ {"num_eq": 10, "num_range": 90, "distinct_range": 10, "upper_bound": "100"},
+ {"num_eq": 10, "num_range": 190, "distinct_range": 10, "upper_bound": "200"},
+ {"num_eq": 20, "num_range": 280, "distinct_range": 10, "upper_bound": "300"},
+ {"num_eq": 30, "num_range": 370, "distinct_range": 10, "upper_bound": "400"}
+ ]
+ },
+ {
+ "columns": ["s"],
+ "created_at": "2018-01-01 1:00:00.00000+00:00",
+ "row_count": 1000,
+ "distinct_count": 40,
+ "null_count": 0,
+ "avg_size": 3,
+ "histo_col_type": "string",
+ "histo_buckets": [
+ {"num_eq": 0, "num_range": 0, "distinct_range": 0, "upper_bound": "apple"},
+ {"num_eq": 100, "num_range": 100, "distinct_range": 10, "upper_bound": "banana"},
+ {"num_eq": 100, "num_range": 100, "distinct_range": 10, "upper_bound": "cherry"},
+ {"num_eq": 200, "num_range": 100, "distinct_range": 10, "upper_bound": "mango"},
+ {"num_eq": 200, "num_range": 100, "distinct_range": 10, "upper_bound": "pineapple"}
+ ]
+ }
+]'
+----
+
+# --------------------------------------------------
+# Q1
+# --------------------------------------------------
+
+opt set=(optimizer_prefer_bounded_cardinality=false)
+SELECT * FROM t WHERE k IN (110, 120, 130, 140) AND i = 500
+----
+index-join t
+ ├── columns: k:1!null i:2!null s:3
+ ├── cardinality: [0 - 4]
+ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0, distinct(1,2)=2e-07, null(1,2)=0]
+ │ histogram(1)= 0 5e-08 0 5e-08 0 5e-08 0 5e-08
+ │ <--- 110 --- 120 --- 130 --- 140
+ │ histogram(2)=
+ ├── cost: 24.0200012
+ ├── key: (1)
+ ├── fd: ()-->(2), (1)-->(3)
+ └── scan t@t_i_idx
+ ├── columns: k:1!null i:2!null
+ ├── constraint: /2/1
+ │ ├── [/500/110 - /500/110]
+ │ ├── [/500/120 - /500/120]
+ │ ├── [/500/130 - /500/130]
+ │ └── [/500/140 - /500/140]
+ ├── cardinality: [0 - 4]
+ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0, distinct(1,2)=2e-07, null(1,2)=0]
+ │ histogram(1)= 0 5e-08 0 5e-08 0 5e-08 0 5e-08
+ │ <--- 110 --- 120 --- 130 --- 140
+ │ histogram(2)=
+ ├── cost: 24.01
+ ├── key: (1)
+ └── fd: ()-->(2)
+
+opt set=(optimizer_prefer_bounded_cardinality=true)
+SELECT * FROM t WHERE k IN (110, 120, 130, 140) AND i = 500
+----
+index-join t
+ ├── columns: k:1!null i:2!null s:3
+ ├── cardinality: [0 - 4]
+ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0, distinct(1,2)=2e-07, null(1,2)=0]
+ │ histogram(1)= 0 5e-08 0 5e-08 0 5e-08 0 5e-08
+ │ <--- 110 --- 120 --- 130 --- 140
+ │ histogram(2)=
+ ├── cost: 24.0200012
+ ├── key: (1)
+ ├── fd: ()-->(2), (1)-->(3)
+ └── scan t@t_i_idx
+ ├── columns: k:1!null i:2!null
+ ├── constraint: /2/1
+ │ ├── [/500/110 - /500/110]
+ │ ├── [/500/120 - /500/120]
+ │ ├── [/500/130 - /500/130]
+ │ └── [/500/140 - /500/140]
+ ├── cardinality: [0 - 4]
+ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0, distinct(1,2)=2e-07, null(1,2)=0]
+ │ histogram(1)= 0 5e-08 0 5e-08 0 5e-08 0 5e-08
+ │ <--- 110 --- 120 --- 130 --- 140
+ │ histogram(2)=
+ ├── cost: 24.01
+ ├── key: (1)
+ └── fd: ()-->(2)
+
+# --------------------------------------------------
+# Q2
+# --------------------------------------------------
+
+opt set=(optimizer_prefer_bounded_cardinality=false)
+SELECT * FROM t WHERE k IN (100, 110, 120, 130) AND i > 500
+----
+index-join t
+ ├── columns: k:1!null i:2!null s:3
+ ├── cardinality: [0 - 4]
+ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0, distinct(1,2)=2e-07, null(1,2)=0]
+ │ histogram(1)= 0 5e-08 0 5e-08 0 5e-08 0 5e-08
+ │ <--- 100 --- 110 --- 120 --- 130
+ │ histogram(2)=
+ ├── cost: 18.0500006
+ ├── key: (1)
+ ├── fd: (1)-->(2,3)
+ └── select
+ ├── columns: k:1!null i:2!null
+ ├── cardinality: [0 - 4]
+ ├── stats: [rows=8e-10, distinct(1)=8e-10, null(1)=0]
+ │ histogram(1)= 0 2e-10 0 2e-10 0 2e-10 0 2e-10
+ │ <--- 100 --- 110 --- 120 --- 130
+ ├── cost: 18.0400002
+ ├── key: (1)
+ ├── fd: (1)-->(2)
+ ├── scan t@t_i_idx
+ │ ├── columns: k:1!null i:2!null
+ │ ├── constraint: /2/1: [/501/100 - ]
+ │ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0]
+ │ │ histogram(1)= 0 0 1.98e-08 2e-10 3.98e-08 2e-10 5.98e-08 2e-10 7.98e-08 2e-10
+ │ │ <--- 0 ---------- 100 ---------- 200 ---------- 300 ---------- 400
+ │ │ histogram(2)=
+ │ ├── cost: 18.0200002
+ │ ├── key: (1)
+ │ └── fd: (1)-->(2)
+ └── filters
+ └── k:1 IN (100, 110, 120, 130) [outer=(1), constraints=(/1: [/100 - /100] [/110 - /110] [/120 - /120] [/130 - /130]; tight)]
+
+opt set=(optimizer_prefer_bounded_cardinality=true)
+SELECT * FROM t WHERE k IN (100, 110, 120, 130) AND i > 500
+----
+select
+ ├── columns: k:1!null i:2!null s:3
+ ├── cardinality: [0 - 4]
+ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0, distinct(1,2)=2e-07, null(1,2)=0]
+ │ histogram(1)= 0 5e-08 0 5e-08 0 5e-08 0 5e-08
+ │ <--- 100 --- 110 --- 120 --- 130
+ │ histogram(2)=
+ ├── cost: 24.21
+ ├── key: (1)
+ ├── fd: (1)-->(2,3)
+ ├── scan t
+ │ ├── columns: k:1!null i:2 s:3
+ │ ├── constraint: /1
+ │ │ ├── [/100 - /100]
+ │ │ ├── [/110 - /110]
+ │ │ ├── [/120 - /120]
+ │ │ └── [/130 - /130]
+ │ ├── cardinality: [0 - 4]
+ │ ├── stats: [rows=4, distinct(1)=4, null(1)=0]
+ │ │ histogram(1)= 0 1 0 1 0 1 0 1
+ │ │ <--- 100 --- 110 --- 120 --- 130
+ │ ├── cost: 24.15
+ │ ├── key: (1)
+ │ └── fd: (1)-->(2,3)
+ └── filters
+ └── i:2 > 500 [outer=(2), constraints=(/2: [/501 - ]; tight)]
+
+# --------------------------------------------------
+# Q3
+# --------------------------------------------------
+
+opt set=(optimizer_prefer_bounded_cardinality=false)
+SELECT * FROM t WHERE k IN (410, 420, 430) AND i > 500
+----
+index-join t
+ ├── columns: k:1!null i:2!null s:3
+ ├── cardinality: [0 - 3]
+ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0]
+ │ histogram(1)=
+ │ histogram(2)=
+ ├── cost: 18.0500006
+ ├── key: (1)
+ ├── fd: (1)-->(2,3)
+ └── select
+ ├── columns: k:1!null i:2!null
+ ├── cardinality: [0 - 3]
+ ├── stats: [rows=4e-17, distinct(1)=4e-17, null(1)=0]
+ │ histogram(1)=
+ ├── cost: 18.0400002
+ ├── key: (1)
+ ├── fd: (1)-->(2)
+ ├── scan t@t_i_idx
+ │ ├── columns: k:1!null i:2!null
+ │ ├── constraint: /2/1: [/501/410 - ]
+ │ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0]
+ │ │ histogram(1)= 0 0 1.98e-08 2e-10 3.98e-08 2e-10 5.98e-08 2e-10 7.98e-08 2e-10
+ │ │ <--- 0 ---------- 100 ---------- 200 ---------- 300 ---------- 400
+ │ │ histogram(2)=
+ │ ├── cost: 18.0200002
+ │ ├── key: (1)
+ │ └── fd: (1)-->(2)
+ └── filters
+ └── k:1 IN (410, 420, 430) [outer=(1), constraints=(/1: [/410 - /410] [/420 - /420] [/430 - /430]; tight)]
+
+opt set=(optimizer_prefer_bounded_cardinality=true)
+SELECT * FROM t WHERE k IN (410, 420, 430) AND i > 500
+----
+select
+ ├── columns: k:1!null i:2!null s:3
+ ├── cardinality: [0 - 3]
+ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0]
+ │ histogram(1)=
+ │ histogram(2)=
+ ├── cost: 19.03
+ ├── key: (1)
+ ├── fd: (1)-->(2,3)
+ ├── scan t
+ │ ├── columns: k:1!null i:2 s:3
+ │ ├── constraint: /1
+ │ │ ├── [/410 - /410]
+ │ │ ├── [/420 - /420]
+ │ │ └── [/430 - /430]
+ │ ├── cardinality: [0 - 3]
+ │ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0]
+ │ │ histogram(1)=
+ │ ├── cost: 19.01
+ │ ├── key: (1)
+ │ └── fd: (1)-->(2,3)
+ └── filters
+ └── i:2 > 500 [outer=(2), constraints=(/2: [/501 - ]; tight)]
+
+# --------------------------------------------------
+# Q4
+# --------------------------------------------------
+
+opt set=(optimizer_prefer_bounded_cardinality=false)
+SELECT * FROM t WHERE k IN (100, 110, 120, 130) AND i = 400 AND s < 'apple'
+----
+select
+ ├── columns: k:1!null i:2!null s:3!null
+ ├── cardinality: [0 - 4]
+ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0, distinct(3)=2e-07, null(3)=0, distinct(2,3)=2e-07, null(2,3)=0, distinct(1-3)=2e-07, null(1-3)=0]
+ │ histogram(1)= 0 5e-08 0 5e-08 0 5e-08 0 5e-08
+ │ <--- 100 --- 110 --- 120 --- 130
+ │ histogram(2)= 0 2e-07
+ │ <--- 400
+ │ histogram(3)=
+ ├── cost: 18.0700002
+ ├── key: (1)
+ ├── fd: ()-->(2), (1)-->(3)
+ ├── index-join t
+ │ ├── columns: k:1!null i:2 s:3
+ │ ├── cardinality: [0 - 4]
+ │ ├── stats: [rows=8e-10]
+ │ ├── cost: 18.0500002
+ │ ├── key: (1)
+ │ ├── fd: (1)-->(2,3)
+ │ └── select
+ │ ├── columns: k:1!null s:3!null
+ │ ├── cardinality: [0 - 4]
+ │ ├── stats: [rows=8e-10, distinct(1)=8e-10, null(1)=0]
+ │ │ histogram(1)= 0 2e-10 0 2e-10 0 2e-10 0 2e-10
+ │ │ <--- 100 --- 110 --- 120 --- 130
+ │ ├── cost: 18.0400002
+ │ ├── key: (1)
+ │ ├── fd: (1)-->(3)
+ │ ├── scan t@t_s_idx
+ │ │ ├── columns: k:1!null s:3!null
+ │ │ ├── constraint: /3/1: (/NULL - /'apple')
+ │ │ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(3)=2e-07, null(3)=0]
+ │ │ │ histogram(1)= 0 0 1.98e-08 2e-10 3.98e-08 2e-10 5.98e-08 2e-10 7.98e-08 2e-10
+ │ │ │ <--- 0 ---------- 100 ---------- 200 ---------- 300 ---------- 400
+ │ │ │ histogram(3)=
+ │ │ ├── cost: 18.0200002
+ │ │ ├── key: (1)
+ │ │ └── fd: (1)-->(3)
+ │ └── filters
+ │ └── k:1 IN (100, 110, 120, 130) [outer=(1), constraints=(/1: [/100 - /100] [/110 - /110] [/120 - /120] [/130 - /130]; tight)]
+ └── filters
+ └── i:2 = 400 [outer=(2), constraints=(/2: [/400 - /400]; tight), fd=()-->(2)]
+
+opt set=(optimizer_prefer_bounded_cardinality=true)
+SELECT * FROM t WHERE k IN (100, 110, 120, 130) AND i = 400 AND s < 'apple'
+----
+select
+ ├── columns: k:1!null i:2!null s:3!null
+ ├── cardinality: [0 - 4]
+ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0, distinct(3)=2e-07, null(3)=0, distinct(2,3)=2e-07, null(2,3)=0, distinct(1-3)=2e-07, null(1-3)=0]
+ │ histogram(1)= 0 5e-08 0 5e-08 0 5e-08 0 5e-08
+ │ <--- 100 --- 110 --- 120 --- 130
+ │ histogram(2)= 0 2e-07
+ │ <--- 400
+ │ histogram(3)=
+ ├── cost: 24.22
+ ├── key: (1)
+ ├── fd: ()-->(2), (1)-->(3)
+ ├── scan t
+ │ ├── columns: k:1!null i:2 s:3
+ │ ├── constraint: /1
+ │ │ ├── [/100 - /100]
+ │ │ ├── [/110 - /110]
+ │ │ ├── [/120 - /120]
+ │ │ └── [/130 - /130]
+ │ ├── cardinality: [0 - 4]
+ │ ├── stats: [rows=4, distinct(1)=4, null(1)=0]
+ │ │ histogram(1)= 0 1 0 1 0 1 0 1
+ │ │ <--- 100 --- 110 --- 120 --- 130
+ │ ├── cost: 24.15
+ │ ├── key: (1)
+ │ └── fd: (1)-->(2,3)
+ └── filters
+ ├── i:2 = 400 [outer=(2), constraints=(/2: [/400 - /400]; tight), fd=()-->(2)]
+ └── s:3 < 'apple' [outer=(3), constraints=(/3: (/NULL - /'apple'); tight)]
+
+# --------------------------------------------------
+# Q5
+# --------------------------------------------------
+
+opt set=(enable_zigzag_join=false,optimizer_prefer_bounded_cardinality=false)
+SELECT * FROM t WHERE i = 400 AND s > 'z'
+----
+select
+ ├── columns: k:1!null i:2!null s:3!null
+ ├── stats: [rows=2e-07, distinct(2)=2e-07, null(2)=0, distinct(3)=2e-07, null(3)=0, distinct(2,3)=2e-07, null(2,3)=0]
+ │ histogram(2)= 0 2e-07
+ │ <--- 400
+ │ histogram(3)=
+ ├── cost: 18.0700014
+ ├── key: (1)
+ ├── fd: ()-->(2), (1)-->(3)
+ ├── index-join t
+ │ ├── columns: k:1!null i:2 s:3
+ │ ├── stats: [rows=2e-07]
+ │ ├── cost: 18.0400014
+ │ ├── key: (1)
+ │ ├── fd: (1)-->(2,3)
+ │ └── scan t@t_s_idx
+ │ ├── columns: k:1!null s:3!null
+ │ ├── constraint: /3/1: [/e'z\x00' - ]
+ │ ├── stats: [rows=2e-07, distinct(3)=2e-07, null(3)=0]
+ │ │ histogram(3)=
+ │ ├── cost: 18.0200002
+ │ ├── key: (1)
+ │ └── fd: (1)-->(3)
+ └── filters
+ └── i:2 = 400 [outer=(2), constraints=(/2: [/400 - /400]; tight), fd=()-->(2)]
+
+opt set=(enable_zigzag_join=false,optimizer_prefer_bounded_cardinality=true)
+SELECT * FROM t WHERE i = 400 AND s > 'z'
+----
+select
+ ├── columns: k:1!null i:2!null s:3!null
+ ├── stats: [rows=2e-07, distinct(2)=2e-07, null(2)=0, distinct(3)=2e-07, null(3)=0, distinct(2,3)=2e-07, null(2,3)=0]
+ │ histogram(2)= 0 2e-07
+ │ <--- 400
+ │ histogram(3)=
+ ├── cost: 18.0700014
+ ├── cost-flags: unbounded-cardinality
+ ├── key: (1)
+ ├── fd: ()-->(2), (1)-->(3)
+ ├── index-join t
+ │ ├── columns: k:1!null i:2 s:3
+ │ ├── stats: [rows=2e-07]
+ │ ├── cost: 18.0400014
+ │ ├── cost-flags: unbounded-cardinality
+ │ ├── key: (1)
+ │ ├── fd: (1)-->(2,3)
+ │ └── scan t@t_s_idx
+ │ ├── columns: k:1!null s:3!null
+ │ ├── constraint: /3/1: [/e'z\x00' - ]
+ │ ├── stats: [rows=2e-07, distinct(3)=2e-07, null(3)=0]
+ │ │ histogram(3)=
+ │ ├── cost: 18.0200002
+ │ ├── cost-flags: unbounded-cardinality
+ │ ├── key: (1)
+ │ └── fd: (1)-->(3)
+ └── filters
+ └── i:2 = 400 [outer=(2), constraints=(/2: [/400 - /400]; tight), fd=()-->(2)]
+
+# --------------------------------------------------
+# Q6
+# --------------------------------------------------
+
+opt set=(optimizer_prefer_bounded_cardinality=false)
+SELECT * FROM t WHERE k = 100 AND i = 500 AND s = 'zzz'
+----
+select
+ ├── columns: k:1!null i:2!null s:3!null
+ ├── cardinality: [0 - 1]
+ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0, distinct(3)=2e-07, null(3)=0]
+ │ histogram(1)= 0 2e-07
+ │ <--- 100
+ │ histogram(2)=
+ │ histogram(3)=
+ ├── cost: 9.04000121
+ ├── key: ()
+ ├── fd: ()-->(1-3)
+ ├── index-join t
+ │ ├── columns: k:1!null i:2 s:3
+ │ ├── cardinality: [0 - 1]
+ │ ├── stats: [rows=2e-07]
+ │ ├── cost: 9.02000121
+ │ ├── key: ()
+ │ ├── fd: ()-->(1-3)
+ │ └── scan t@t_i_idx
+ │ ├── columns: k:1!null i:2!null
+ │ ├── constraint: /2/1: [/500/100 - /500/100]
+ │ ├── cardinality: [0 - 1]
+ │ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0, distinct(1,2)=2e-07, null(1,2)=0]
+ │ │ histogram(1)= 0 2e-07
+ │ │ <--- 100
+ │ │ histogram(2)=
+ │ ├── cost: 9.01
+ │ ├── key: ()
+ │ └── fd: ()-->(1,2)
+ └── filters
+ └── s:3 = 'zzz' [outer=(3), constraints=(/3: [/'zzz' - /'zzz']; tight), fd=()-->(3)]
+
+opt set=(optimizer_prefer_bounded_cardinality=true)
+SELECT * FROM t WHERE k = 100 AND i = 500 AND s = 'zzz'
+----
+select
+ ├── columns: k:1!null i:2!null s:3!null
+ ├── cardinality: [0 - 1]
+ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0, distinct(3)=2e-07, null(3)=0]
+ │ histogram(1)= 0 2e-07
+ │ <--- 100
+ │ histogram(2)=
+ │ histogram(3)=
+ ├── cost: 9.04000121
+ ├── key: ()
+ ├── fd: ()-->(1-3)
+ ├── index-join t
+ │ ├── columns: k:1!null i:2 s:3
+ │ ├── cardinality: [0 - 1]
+ │ ├── stats: [rows=2e-07]
+ │ ├── cost: 9.02000121
+ │ ├── key: ()
+ │ ├── fd: ()-->(1-3)
+ │ └── scan t@t_i_idx
+ │ ├── columns: k:1!null i:2!null
+ │ ├── constraint: /2/1: [/500/100 - /500/100]
+ │ ├── cardinality: [0 - 1]
+ │ ├── stats: [rows=2e-07, distinct(1)=2e-07, null(1)=0, distinct(2)=2e-07, null(2)=0, distinct(1,2)=2e-07, null(1,2)=0]
+ │ │ histogram(1)= 0 2e-07
+ │ │ <--- 100
+ │ │ histogram(2)=
+ │ ├── cost: 9.01
+ │ ├── key: ()
+ │ └── fd: ()-->(1,2)
+ └── filters
+ └── s:3 = 'zzz' [outer=(3), constraints=(/3: [/'zzz' - /'zzz']; tight), fd=()-->(3)]
diff --git a/pkg/sql/prepared_stmt.go b/pkg/sql/prepared_stmt.go
index 967fb36b487a..681222ca6122 100644
--- a/pkg/sql/prepared_stmt.go
+++ b/pkg/sql/prepared_stmt.go
@@ -164,6 +164,11 @@ func (p *planCosts) NumCustom() int {
// AvgCustom returns the average cost of all the custom plan costs in planCosts.
// If there are no custom plan costs, it returns 0.
+//
+// TODO(mgartner): Figure out how this should incorporate cost flags. Some of
+// them, like UnboundedCardinality, are only set if session settings are set.
+// When those session settings change, do we need to clear and recompute the
+// average cost of custom plans?
func (p *planCosts) AvgCustom() memo.Cost {
if p.custom.length == 0 {
return memo.Cost{C: 0}
diff --git a/pkg/sql/sessiondatapb/local_only_session_data.proto b/pkg/sql/sessiondatapb/local_only_session_data.proto
index f07b3c86e0da..63694ead622e 100644
--- a/pkg/sql/sessiondatapb/local_only_session_data.proto
+++ b/pkg/sql/sessiondatapb/local_only_session_data.proto
@@ -594,6 +594,10 @@ message LocalOnlySessionData {
// CatalogDigestStalenessCheckEnabled is used to enable using the catalog
// digest information to do fast memo checks.
bool catalog_digest_staleness_check_enabled = 153;
+ // OptimizerPreferBoundedCardinality instructs the optimizer to prefer query
+ // plans in which every expression has a bounded cardinality over plans with
+ // one or more expressions with unbounded cardinality.
+ bool optimizer_prefer_bounded_cardinality = 154;
///////////////////////////////////////////////////////////////////////////
// WARNING: consider whether a session parameter you're adding needs to //
diff --git a/pkg/sql/vars.go b/pkg/sql/vars.go
index b4456217ca82..b85c2aed493d 100644
--- a/pkg/sql/vars.go
+++ b/pkg/sql/vars.go
@@ -3724,6 +3724,7 @@ var varGen = map[string]sessionVar{
},
GlobalDefault: globalFalse,
},
+
// CockroachDB extension.
`catalog_digest_staleness_check_enabled`: {
GetStringVal: makePostgresBoolGetStringValFn(`catalog_digest_staleness_check_enabled`),
@@ -3741,6 +3742,23 @@ var varGen = map[string]sessionVar{
GlobalDefault: globalTrue,
Hidden: true,
},
+
+ // CockroachDB extension.
+ `optimizer_prefer_bounded_cardinality`: {
+ GetStringVal: makePostgresBoolGetStringValFn(`optimizer_prefer_bounded_cardinality`),
+ Set: func(_ context.Context, m sessionDataMutator, s string) error {
+ b, err := paramparse.ParseBoolVar("optimizer_prefer_bounded_cardinality", s)
+ if err != nil {
+ return err
+ }
+ m.SetOptimizerPreferBoundedCardinality(b)
+ return nil
+ },
+ Get: func(evalCtx *extendedEvalContext, _ *kv.Txn) (string, error) {
+ return formatBoolAsPostgresSetting(evalCtx.SessionData().OptimizerPreferBoundedCardinality), nil
+ },
+ GlobalDefault: globalFalse,
+ },
}
func ReplicationModeFromString(s string) (sessiondatapb.ReplicationMode, error) {