From 129d9bc99dd8a7da40e2b51cdf45a44eca0e53e0 Mon Sep 17 00:00:00 2001
From: Yuanjia Zhang <zhangyuanjia@pingcap.com>
Date: Wed, 15 Jun 2022 16:38:34 +0800
Subject: [PATCH] planner: introduce new cost formula for
 Selection/TableScan/IndexScan (#35378)

ref pingcap/tidb#35240
---
 planner/core/plan_cost.go | 100 ++++++++++++++++++++++++++++++--------
 1 file changed, 80 insertions(+), 20 deletions(-)

diff --git a/planner/core/plan_cost.go b/planner/core/plan_cost.go
index bfa04ad7a5803..3ec446369631a 100644
--- a/planner/core/plan_cost.go
+++ b/planner/core/plan_cost.go
@@ -34,6 +34,11 @@ const (
 	CostFlagUseTrueCardinality
 )
 
+const (
+	modelVer1 = 1
+	modelVer2 = 2
+)
+
 func hasCostFlag(costFlag, flag uint64) bool {
 	return (costFlag & flag) > 0
 }
@@ -61,21 +66,40 @@ func (p *PhysicalSelection) GetPlanCost(taskType property.TaskType, costFlag uin
 	if p.planCostInit && !hasCostFlag(costFlag, CostFlagRecalculate) {
 		return p.planCost, nil
 	}
-	var cpuFactor float64
-	switch taskType {
-	case property.RootTaskType, property.MppTaskType:
-		cpuFactor = p.ctx.GetSessionVars().GetCPUFactor()
-	case property.CopSingleReadTaskType, property.CopDoubleReadTaskType:
-		cpuFactor = p.ctx.GetSessionVars().GetCopCPUFactor()
-	default:
-		return 0, errors.Errorf("unknown task type %v", taskType)
+
+	var selfCost float64
+	switch p.ctx.GetSessionVars().CostModelVersion {
+	case modelVer1: // selection cost: rows * cpu-factor
+		var cpuFactor float64
+		switch taskType {
+		case property.RootTaskType, property.MppTaskType:
+			cpuFactor = p.ctx.GetSessionVars().GetCPUFactor()
+		case property.CopSingleReadTaskType, property.CopDoubleReadTaskType:
+			cpuFactor = p.ctx.GetSessionVars().GetCopCPUFactor()
+		default:
+			return 0, errors.Errorf("unknown task type %v", taskType)
+		}
+		selfCost = getCardinality(p.children[0], costFlag) * cpuFactor
+	case modelVer2: // selection cost: rows * num-filters * cpu-factor
+		var cpuFactor float64
+		switch taskType {
+		case property.RootTaskType:
+			cpuFactor = p.ctx.GetSessionVars().GetCPUFactor()
+		case property.MppTaskType: // use a dedicated cpu-factor for TiFlash
+			cpuFactor = p.ctx.GetSessionVars().GetTiFlashCPUFactor()
+		case property.CopSingleReadTaskType, property.CopDoubleReadTaskType:
+			cpuFactor = p.ctx.GetSessionVars().GetCopCPUFactor()
+		default:
+			return 0, errors.Errorf("unknown task type %v", taskType)
+		}
+		selfCost = getCardinality(p.children[0], costFlag) * float64(len(p.Conditions)) * cpuFactor
 	}
+
 	childCost, err := p.children[0].GetPlanCost(taskType, costFlag)
 	if err != nil {
 		return 0, err
 	}
-	p.planCost = childCost
-	p.planCost += getCardinality(p.children[0], costFlag) * cpuFactor // selection cost: rows * cpu-factor
+	p.planCost = childCost + selfCost
 	p.planCostInit = true
 	return p.planCost, nil
 }
@@ -343,12 +367,34 @@ func (p *PhysicalTableScan) GetPlanCost(taskType property.TaskType, costFlag uin
 	if p.planCostInit && !hasCostFlag(costFlag, CostFlagRecalculate) {
 		return p.planCost, nil
 	}
-	// scan cost: rows * row-size * scan-factor
-	scanFactor := p.ctx.GetSessionVars().GetScanFactor(p.Table)
-	if p.Desc {
-		scanFactor = p.ctx.GetSessionVars().GetDescScanFactor(p.Table)
+
+	var selfCost float64
+	switch p.ctx.GetSessionVars().CostModelVersion {
+	case modelVer1: // scan cost: rows * row-size * scan-factor
+		scanFactor := p.ctx.GetSessionVars().GetScanFactor(p.Table)
+		if p.Desc {
+			scanFactor = p.ctx.GetSessionVars().GetDescScanFactor(p.Table)
+		}
+		selfCost = getCardinality(p, costFlag) * p.getScanRowSize() * scanFactor
+	case modelVer2: // scan cost: rows * log2(row-size) * scan-factor
+		var scanFactor float64
+		switch taskType {
+		case property.MppTaskType: // use a dedicated scan-factor for TiFlash
+			// no need to distinguish `Scan` and `DescScan` for TiFlash for now
+			scanFactor = p.ctx.GetSessionVars().GetTiFlashScanFactor()
+		default: // for TiKV
+			scanFactor = p.ctx.GetSessionVars().GetScanFactor(p.Table)
+			if p.Desc {
+				scanFactor = p.ctx.GetSessionVars().GetDescScanFactor(p.Table)
+			}
+		}
+		// the formula `log(rowSize)` is based on experiment results
+		rowSize := math.Max(p.getScanRowSize(), 2.0) // to guarantee logRowSize >= 1
+		logRowSize := math.Log2(rowSize)
+		selfCost = getCardinality(p, costFlag) * logRowSize * scanFactor
 	}
-	p.planCost = getCardinality(p, costFlag) * p.getScanRowSize() * scanFactor
+
+	p.planCost = selfCost
 	p.planCostInit = true
 	return p.planCost, nil
 }
@@ -358,12 +404,26 @@ func (p *PhysicalIndexScan) GetPlanCost(taskType property.TaskType, costFlag uin
 	if p.planCostInit && !hasCostFlag(costFlag, CostFlagRecalculate) {
 		return p.planCost, nil
 	}
-	// scan cost: rows * row-size * scan-factor
-	scanFactor := p.ctx.GetSessionVars().GetScanFactor(p.Table)
-	if p.Desc {
-		scanFactor = p.ctx.GetSessionVars().GetDescScanFactor(p.Table)
+
+	var selfCost float64
+	switch p.ctx.GetSessionVars().CostModelVersion {
+	case modelVer1: // scan cost: rows * row-size * scan-factor
+		scanFactor := p.ctx.GetSessionVars().GetScanFactor(p.Table)
+		if p.Desc {
+			scanFactor = p.ctx.GetSessionVars().GetDescScanFactor(p.Table)
+		}
+		selfCost = getCardinality(p, costFlag) * p.getScanRowSize() * scanFactor
+	case modelVer2:
+		scanFactor := p.ctx.GetSessionVars().GetScanFactor(p.Table)
+		if p.Desc {
+			scanFactor = p.ctx.GetSessionVars().GetDescScanFactor(p.Table)
+		}
+		rowSize := math.Max(p.getScanRowSize(), 2.0)
+		logRowSize := math.Log2(rowSize)
+		selfCost = getCardinality(p, costFlag) * logRowSize * scanFactor
 	}
-	p.planCost = getCardinality(p, costFlag) * p.getScanRowSize() * scanFactor
+
+	p.planCost = selfCost
 	p.planCostInit = true
 	return p.planCost, nil
 }