From 8454e1f77259dfa790ad393894d1370ca2b9e5ae Mon Sep 17 00:00:00 2001
From: Yuanjia Zhang <zhangyuanjia@pingcap.com>
Date: Thu, 16 Jun 2022 11:04:34 +0800
Subject: [PATCH] planner: introduce new cost formula for IndexLookup (#35408)

ref pingcap/tidb#35240
---
 planner/core/plan_cost.go | 28 +++++++++++++++++++++++-----
 1 file changed, 23 insertions(+), 5 deletions(-)

diff --git a/planner/core/plan_cost.go b/planner/core/plan_cost.go
index 61308dd8fb3db..386a93505ae30 100644
--- a/planner/core/plan_cost.go
+++ b/planner/core/plan_cost.go
@@ -199,12 +199,14 @@ func (p *PhysicalIndexLookUpReader) GetPlanCost(taskType property.TaskType, cost
 	for tmp = p.tablePlan; len(tmp.Children()) > 0; tmp = tmp.Children()[0] {
 	}
 	ts := tmp.(*PhysicalTableScan)
-	tblCost, err := ts.GetPlanCost(property.CopDoubleReadTaskType, costFlag)
-	if err != nil {
-		return 0, err
+	if p.ctx.GetSessionVars().CostModelVersion == modelVer1 {
+		tblCost, err := ts.GetPlanCost(property.CopDoubleReadTaskType, costFlag)
+		if err != nil {
+			return 0, err
+		}
+		p.planCost -= tblCost
+		p.planCost += getCardinality(p.indexPlan, costFlag) * ts.getScanRowSize() * p.SCtx().GetSessionVars().GetScanFactor(ts.Table)
 	}
-	p.planCost -= tblCost
-	p.planCost += getCardinality(p.indexPlan, costFlag) * ts.getScanRowSize() * p.SCtx().GetSessionVars().GetScanFactor(ts.Table)
 
 	// index-side net I/O cost: rows * row-size * net-factor
 	netFactor := getTableNetFactor(p.tablePlan)
@@ -221,6 +223,12 @@ func (p *PhysicalIndexLookUpReader) GetPlanCost(taskType property.TaskType, cost
 	// table-side seek cost
 	p.planCost += estimateNetSeekCost(p.tablePlan)
 
+	if p.ctx.GetSessionVars().CostModelVersion == modelVer2 {
+		// accumulate the real double-read cost: numDoubleReadTasks * seekFactor
+		numDoubleReadTasks := p.estNumDoubleReadTasks(costFlag)
+		p.planCost += numDoubleReadTasks * p.ctx.GetSessionVars().GetSeekFactor(ts.Table)
+	}
+
 	// consider concurrency
 	p.planCost /= float64(p.ctx.GetSessionVars().DistSQLScanConcurrency())
 
@@ -230,6 +238,16 @@ func (p *PhysicalIndexLookUpReader) GetPlanCost(taskType property.TaskType, cost
 	return p.planCost, nil
 }
 
+func (p *PhysicalIndexLookUpReader) estNumDoubleReadTasks(costFlag uint64) float64 {
+	doubleReadRows := p.indexPlan.StatsCount()
+	batchSize := float64(p.ctx.GetSessionVars().IndexLookupSize)
+	// distRatio indicates how many requests corresponding to a batch, current value is from experiments.
+	// TODO: estimate it by using index correlation or make it configurable.
+	distRatio := 40.0
+	numDoubleReadTasks := (doubleReadRows / batchSize) * distRatio
+	return numDoubleReadTasks // use Float64 instead of Int like `Ceil(...)` to make the cost continuous
+}
+
 // GetPlanCost calculates the cost of the plan if it has not been calculated yet and returns the cost.
 func (p *PhysicalIndexReader) GetPlanCost(taskType property.TaskType, costFlag uint64) (float64, error) {
 	if p.planCostInit && !hasCostFlag(costFlag, CostFlagRecalculate) {