planner: refactor the cost implementation for PointGet/BatchGet (ping…

…cap#34300)

ref pingcap#33945

planner/core/find_best_task.go

@@ -1964,7 +1964,8 @@ func (ds *DataSource) convertToPointGet(prop *property.PhysicalProperty, candida
 		pointGetPlan.Handle = kv.IntHandle(candidate.path.Ranges[0].LowVal[0].GetInt64())
 		pointGetPlan.UnsignedHandle = mysql.HasUnsignedFlag(ds.handleCols.GetCol(0).RetType.GetFlag())
 		pointGetPlan.PartitionInfo = partitionInfo
-		cost = pointGetPlan.GetCost(ds.TblCols)
+		pointGetPlan.accessCols = ds.TblCols
+		cost = pointGetPlan.GetCost()
 		// Add filter condition to table plan now.
 		if len(candidate.path.TableFilters) > 0 {
 			sessVars := ds.ctx.GetSessionVars()
@@ -1982,10 +1983,11 @@ func (ds *DataSource) convertToPointGet(prop *property.PhysicalProperty, candida
 		pointGetPlan.IndexValues = candidate.path.Ranges[0].LowVal
 		pointGetPlan.PartitionInfo = partitionInfo
 		if candidate.path.IsSingleScan {
-			cost = pointGetPlan.GetCost(candidate.path.IdxCols)
+			pointGetPlan.accessCols = candidate.path.IdxCols
 		} else {
-			cost = pointGetPlan.GetCost(ds.TblCols)
+			pointGetPlan.accessCols = ds.TblCols
 		}
+		cost = pointGetPlan.GetCost()
 		// Add index condition to table plan now.
 		if len(candidate.path.IndexFilters)+len(candidate.path.TableFilters) > 0 {
 			sessVars := ds.ctx.GetSessionVars()
@@ -1999,7 +2001,7 @@ func (ds *DataSource) convertToPointGet(prop *property.PhysicalProperty, candida
 	}
 
 	rTsk.cst = cost
-	pointGetPlan.SetCost(cost)
+	rTsk.p.SetCost(cost)
 	return rTsk
 }
 
@@ -2033,7 +2035,8 @@ func (ds *DataSource) convertToBatchPointGet(prop *property.PhysicalProperty,
 		for _, ran := range candidate.path.Ranges {
 			batchPointGetPlan.Handles = append(batchPointGetPlan.Handles, kv.IntHandle(ran.LowVal[0].GetInt64()))
 		}
-		cost = batchPointGetPlan.GetCost(ds.TblCols)
+		batchPointGetPlan.accessCols = ds.TblCols
+		cost = batchPointGetPlan.GetCost()
 		// Add filter condition to table plan now.
 		if len(candidate.path.TableFilters) > 0 {
 			sessVars := ds.ctx.GetSessionVars()
@@ -2057,10 +2060,11 @@ func (ds *DataSource) convertToBatchPointGet(prop *property.PhysicalProperty,
 			batchPointGetPlan.Desc = prop.SortItems[0].Desc
 		}
 		if candidate.path.IsSingleScan {
-			cost = batchPointGetPlan.GetCost(candidate.path.IdxCols)
+			batchPointGetPlan.accessCols = candidate.path.IdxCols
 		} else {
-			cost = batchPointGetPlan.GetCost(ds.TblCols)
+			batchPointGetPlan.accessCols = ds.TblCols
 		}
+		cost = batchPointGetPlan.GetCost()
 		// Add index condition to table plan now.
 		if len(candidate.path.IndexFilters)+len(candidate.path.TableFilters) > 0 {
 			sessVars := ds.ctx.GetSessionVars()
@@ -2074,7 +2078,7 @@ func (ds *DataSource) convertToBatchPointGet(prop *property.PhysicalProperty,
 	}
 
 	rTsk.cst = cost
-	batchPointGetPlan.SetCost(cost)
+	rTsk.p.SetCost(cost)
 	return rTsk
 }
 

planner/core/plan_cost.go

@@ -349,12 +349,22 @@ func (p *PhysicalTopN) GetPlanCost(taskType property.TaskType) (float64, error)
 
 // GetPlanCost calculates the cost of the plan if it has not been calculated yet and returns the cost.
 func (p *BatchPointGetPlan) GetPlanCost(taskType property.TaskType) (float64, error) {
-	return 0, errors.New("not implemented")
+	if p.planCostInit {
+		return p.planCost, nil
+	}
+	p.planCost = p.GetCost()
+	p.planCostInit = true
+	return p.planCost, nil
 }
 
 // GetPlanCost calculates the cost of the plan if it has not been calculated yet and returns the cost.
 func (p *PointGetPlan) GetPlanCost(taskType property.TaskType) (float64, error) {
-	return 0, errors.New("not implemented")
+	if p.planCostInit {
+		return p.planCost, nil
+	}
+	p.planCost = p.GetCost()
+	p.planCostInit = true
+	return p.planCost, nil
 }
 
 // GetPlanCost calculates the cost of the plan if it has not been calculated yet and returns the cost.

planner/core/plan_cost_test.go

@@ -48,7 +48,7 @@ func TestNewCostInterfaceTiKV(t *testing.T) {
 	tk := testkit.NewTestKit(t, store)
 
 	tk.MustExec("use test")
-	tk.MustExec(`create table t (a int primary key, b int, c int, d int, key b(b), key cd(c, d))`)
+	tk.MustExec(`create table t (a int primary key, b int, c int, d int, k int, key b(b), key cd(c, d), unique key(k))`)
 
 	queries := []string{
 		// table-reader
@@ -193,6 +193,15 @@ func TestNewCostInterfaceTiKV(t *testing.T) {
 		"select * from t use index(cd) where c < 200 order by c limit 10",
 		"select * from t use index(cd) where c = 200 order by c limit 10",
 		"select * from t use index(cd) where c = 200 and d < 200 order by c, d limit 10",
+		// point get
+		"select * from t where a = 1", // generated in fast plan optimization
+		"select * from t where a in (1, 2, 3, 4, 5)",
+		"select * from t where k = 1",
+		"select * from t where k in (1, 2, 3, 4, 5)",
+		"select * from t where a=1 and mod(a, b)=2", // generated in physical plan optimization
+		"select * from t where a in (1, 2, 3, 4, 5) and mod(a, b)=2",
+		"select * from t where k=1 and mod(k, b)=2",
+		"select * from t where k in (1, 2, 3, 4, 5) and mod(k, b)=2",
 		// union all
 		"select * from t use index(primary) union all select * from t use index(primary) where a < 200",
 		"select b from t use index(primary) union all select b from t use index(b) where b < 200",

planner/core/point_get_plan.go

@@ -80,6 +80,12 @@ type PointGetPlan struct {
 	partitionColumnPos int
 	Columns            []*model.ColumnInfo
 	cost               float64
+
+	// required by cost model
+	planCostInit bool
+	planCost     float64
+	// accessCols represents actual columns the PointGet will access, which are used to calculate row-size
+	accessCols []*expression.Column
 }
 
 type nameValuePair struct {
@@ -256,7 +262,11 @@ func (p *PointGetPlan) SetOutputNames(names types.NameSlice) {
 }
 
 // GetCost returns cost of the PointGetPlan.
-func (p *PointGetPlan) GetCost(cols []*expression.Column) float64 {
+func (p *PointGetPlan) GetCost() float64 {
+	cols := p.accessCols
+	if cols == nil {
+		return 0 // the cost of PointGet generated in fast plan optimization is always 0
+	}
 	sessVars := p.ctx.GetSessionVars()
 	var rowSize float64
 	cost := 0.0
@@ -305,6 +315,12 @@ type BatchPointGetPlan struct {
 	SinglePart bool
 	// PartTblID is the table ID for the specific table partition.
 	PartTblID int64
+
+	// required by cost model
+	planCostInit bool
+	planCost     float64
+	// accessCols represents actual columns the PointGet will access, which are used to calculate row-size
+	accessCols []*expression.Column
 }
 
 // Cost implements PhysicalPlan interface
@@ -444,7 +460,11 @@ func (p *BatchPointGetPlan) SetOutputNames(names types.NameSlice) {
 }
 
 // GetCost returns cost of the PointGetPlan.
-func (p *BatchPointGetPlan) GetCost(cols []*expression.Column) float64 {
+func (p *BatchPointGetPlan) GetCost() float64 {
+	cols := p.accessCols
+	if cols == nil {
+		return 0 // the cost of BatchGet generated in fast plan optimization is always 0
+	}
 	sessVars := p.ctx.GetSessionVars()
 	var rowSize, rowCount float64
 	cost := 0.0

planner/core/testdata/integration_suite_out.json

@@ -1723,8 +1723,8 @@
       {
         "SQL": "select a, b, c from t2 where a = 1 and b = 2 and c in (1, 2, 3, 4, 5)",
         "Plan": [
-          "Selection_6 0.01 0.00 root  eq(test.t2.b, 2), in(test.t2.c, 1, 2, 3, 4, 5)",
-          "└─Point_Get_5 1.00 8.60 root table:t2, index:idx_a(a) "
+          "Selection_6 0.01 8.60 root  eq(test.t2.b, 2), in(test.t2.c, 1, 2, 3, 4, 5)",
+          "└─Point_Get_5 1.00 0.00 root table:t2, index:idx_a(a) "
         ],
         "Warnings": [
           "Note 1105 unique index idx_a of t2 is selected since the path only has point ranges with double scan"
@@ -1780,8 +1780,8 @@
         "SQL": "delete from t1 where f = 2 and g in (3, 4)",
         "Plan": [
           "Delete_4 N/A N/A root  N/A",
-          "└─Selection_7 2.00 0.00 root  in(test.t1.g, 3, 4)",
-          "  └─Point_Get_6 1.00 9.80 root table:t1, index:f(f) "
+          "└─Selection_7 2.00 9.80 root  in(test.t1.g, 3, 4)",
+          "  └─Point_Get_6 1.00 0.00 root table:t1, index:f(f) "
         ],
         "Warnings": [
           "Note 1105 unique index f of t1 is selected since the path only has point ranges with double scan"