为什么两个查询中的第一个经过的执行时间更低,而估计的子树成本更高?
估计的子树成本只是与其他查询性能指标(如响应时间)一起衡量查询计划性能的指南,但我对逆的存在感到惊讶并排比较两个相似查询的成本和时间之间的关系。
我已经创建了一个示例数据库来说明这个问题,并为两个查询包含了生成的XML查询计划。我意识到StateID作为一个外键在State和Town表中的存在是不好的形式,但这是我最简单的方式来说明这种现象。
我不只是问“哪个更重要,子树成本或响应时间。”我想了解两个类似查询的查询计划子树成本和响应时间之间如何存在反比关系,这样我就可以做出更明智的决策。
请注意,下面的两个查询会产生完全相同的结果集。
谢谢!
CREATE TABLE dbo.County (StateID INT, CountyID INT)
CREATE TABLE dbo.Town (StateID INT, CountyID INT, TownID INT)
DECLARE @State TABLE (StateID INT, PRIMARY KEY CLUSTERED (StateID))
DECLARE @County TABLE (CountyID INT, PRIMARY KEY CLUSTERED (CountyID))
INSERT INTO @State
VALUES (27)
INSERT INTO @County
SELECT DISTINCT
c.CountyID
FROM dbo.County c
JOIN @State s ON c.StateID = s.StateID
-- low time, high cost query
SELECT DISTINCT
t.CountyID,
t.TownID
FROM dbo.Town t
JOIN @State s
ON t.StateID = s.StateID
-- low cost, high time query
SELECT DISTINCT
t.CountyID,
t.TownID
FROM dbo.Town t
JOIN @State s
ON t.StateID = s.StateID
JOIN @County c
ON t.CountyID = c.CountyID
以下查询计划适用于估计子树成本为6.06且经过时间为1114毫秒的查询。
<StmtSimple StatementCompId="6" StatementEstRows="273495" StatementId="3" StatementOptmLevel="FULL" StatementSubTreeCost="6.06304" StatementText="SELECT DISTINCT
	t.CountyID,
	t.TownID
FROM dbo.Town t
JOIN @State s
	ON t.StateID = s.StateID

-- low cost, high time query
" StatementType="SELECT" QueryHash="0x9347C19165C31DBF" QueryPlanHash="0x98EE57C66D8B347A">
<StatementSetOptions ANSI_NULLS="true" ANSI_PADDING="true" ANSI_WARNINGS="true" ARITHABORT="true" CONCAT_NULL_YIELDS_NULL="true" NUMERIC_ROUNDABORT="false" QUOTED_IDENTIFIER="true" />
<QueryPlan DegreeOfParallelism="2" MemoryGrant="28144" CachedPlanSize="32" CompileTime="8" CompileCPU="8" CompileMemory="312">
<RelOp AvgRowSize="15" EstimateCPU="0.388488" EstimateIO="0" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="273495" LogicalOp="Gather Streams" NodeId="0" Parallel="true" PhysicalOp="Parallelism" EstimatedTotalSubtreeCost="6.06304">
<OutputList>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</OutputList>
<RunTimeInformation>
<RunTimeCountersPerThread Thread="0" ActualRows="13027" ActualEndOfScans="1" ActualExecutions="1" />
</RunTimeInformation>
<Parallelism>
<RelOp AvgRowSize="15" EstimateCPU="3.79725" EstimateIO="0" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="273495" LogicalOp="Aggregate" NodeId="1" Parallel="true" PhysicalOp="Hash Match" EstimatedTotalSubtreeCost="5.67455">
<OutputList>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</OutputList>
<MemoryFractions Input="0.998723" Output="1" />
<RunTimeInformation>
<RunTimeCountersPerThread Thread="2" ActualRows="6515" ActualEndOfScans="1" ActualExecutions="1" />
<RunTimeCountersPerThread Thread="1" ActualRows="6512" ActualEndOfScans="1" ActualExecutions="1" />
<RunTimeCountersPerThread Thread="0" ActualRows="0" ActualEndOfScans="0" ActualExecutions="0" />
</RunTimeInformation>
<Hash>
<DefinedTowns />
<HashKeysBuild>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</HashKeysBuild>
<BuildResidual>
<ScalarOperator ScalarString="[Locality].[dbo].[Town].[CountyID] as [t].[CountyID] = [Locality].[dbo].[Town].[CountyID] as [t].[CountyID] AND [Locality].[dbo].[Town].[TownID] as [t].[TownID] = [Locality].[dbo].[Town].[TownID] as [t].[TownID]">
<Logical Operation="AND">
<ScalarOperator>
<Compare CompareOp="IS">
<ScalarOperator>
<Identifier>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
</Identifier>
</ScalarOperator>
<ScalarOperator>
<Identifier>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
</Identifier>
</ScalarOperator>
</Compare>
</ScalarOperator>
<ScalarOperator>
<Compare CompareOp="IS">
<ScalarOperator>
<Identifier>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</Identifier>
</ScalarOperator>
<ScalarOperator>
<Identifier>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</Identifier>
</ScalarOperator>
</Compare>
</ScalarOperator>
</Logical>
</ScalarOperator>
</BuildResidual>
<RelOp AvgRowSize="15" EstimateCPU="0.494228" EstimateIO="0" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="273495" LogicalOp="Repartition Streams" NodeId="2" Parallel="true" PhysicalOp="Parallelism" EstimatedTotalSubtreeCost="1.8773">
<OutputList>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</OutputList>
<RunTimeInformation>
<RunTimeCountersPerThread Thread="2" ActualRows="1017328" ActualEndOfScans="1" ActualExecutions="1" />
<RunTimeCountersPerThread Thread="1" ActualRows="1093191" ActualEndOfScans="1" ActualExecutions="1" />
<RunTimeCountersPerThread Thread="0" ActualRows="0" ActualEndOfScans="0" ActualExecutions="0" />
</RunTimeInformation>
<Parallelism PartitioningType="Hash">
<PartitionColumns>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</PartitionColumns>
<RelOp AvgRowSize="15" EstimateCPU="0.571604" EstimateIO="0" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="273495" LogicalOp="Inner Join" NodeId="3" Parallel="true" PhysicalOp="Nested Loops" EstimatedTotalSubtreeCost="1.38307">
<OutputList>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</OutputList>
<RunTimeInformation>
<RunTimeCountersPerThread Thread="2" ActualRows="0" ActualEndOfScans="1" ActualExecutions="1" />
<RunTimeCountersPerThread Thread="1" ActualRows="2110519" ActualEndOfScans="1" ActualExecutions="1" />
<RunTimeCountersPerThread Thread="0" ActualRows="0" ActualEndOfScans="0" ActualExecutions="0" />
</RunTimeInformation>
<NestedLoops Optimized="true">
<OuterReferences>
<ColumnReference Table="@State" Alias="[s]" Column="StateID" />
</OuterReferences>
<RelOp AvgRowSize="11" EstimateCPU="0.0285019" EstimateIO="0" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="1" LogicalOp="Distribute Streams" NodeId="5" Parallel="true" PhysicalOp="Parallelism" EstimatedTotalSubtreeCost="0.031785">
<OutputList>
<ColumnReference Table="@State" Alias="[s]" Column="StateID" />
</OutputList>
<RunTimeInformation>
<RunTimeCountersPerThread Thread="2" ActualRows="0" ActualEndOfScans="1" ActualExecutions="1" />
<RunTimeCountersPerThread Thread="1" ActualRows="1" ActualEndOfScans="1" ActualExecutions="1" />
<RunTimeCountersPerThread Thread="0" ActualRows="0" ActualEndOfScans="0" ActualExecutions="0" />
</RunTimeInformation>
<Parallelism PartitioningType="RoundRobin">
<RelOp AvgRowSize="11" EstimateCPU="0.0001581" EstimateIO="0.003125" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="1" LogicalOp="Clustered Index Scan" NodeId="6" Parallel="false" PhysicalOp="Clustered Index Scan" EstimatedTotalSubtreeCost="0.0032831" TableCardinality="0">
<OutputList>
<ColumnReference Table="@State" Alias="[s]" Column="StateID" />
</OutputList>
<RunTimeInformation>
<RunTimeCountersPerThread Thread="1" ActualRows="1" ActualEndOfScans="1" ActualExecutions="1" />
<RunTimeCountersPerThread Thread="0" ActualRows="0" ActualEndOfScans="0" ActualExecutions="0" />
</RunTimeInformation>
<IndexScan Ordered="false" ForcedIndex="false" NoExpandHint="false">
<DefinedTowns>
<DefinedTown>
<ColumnReference Table="@State" Alias="[s]" Column="StateID" />
</DefinedTown>
</DefinedTowns>
<Object Table="[@State]" Index="[PK__#3AB788A__C3BA3B5A3C9FD11A]" Alias="[s]" />
</IndexScan>
</RelOp>
</Parallelism>
</RelOp>
<RelOp AvgRowSize="15" EstimateCPU="0.301001" EstimateIO="0.478681" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="273495" LogicalOp="Index Seek" NodeId="7" Parallel="true" PhysicalOp="Index Seek" EstimatedTotalSubtreeCost="0.779682" TableCardinality="6291320">
<OutputList>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</OutputList>
<RunTimeInformation>
<RunTimeCountersPerThread Thread="2" ActualRows="0" ActualEndOfScans="0" ActualExecutions="0" />
<RunTimeCountersPerThread Thread="1" ActualRows="2110519" ActualEndOfScans="1" ActualExecutions="1" />
<RunTimeCountersPerThread Thread="0" ActualRows="0" ActualEndOfScans="0" ActualExecutions="0" />
</RunTimeInformation>
<IndexScan Ordered="true" ScanDirection="FORWARD" ForcedIndex="false" ForceSeek="false" NoExpandHint="false">
<DefinedTowns>
<DefinedTown>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
</DefinedTown>
<DefinedTown>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</DefinedTown>
</DefinedTowns>
<Object Database="[Locality]" Schema="[dbo]" Table="[Town]" Index="[NCX_Town_StateID_CountyID_TownID]" Alias="[t]" IndexKind="NonClustered" />
<SeekPredicates>
<SeekPredicateNew>
<SeekKeys>
<Prefix ScanType="EQ">
<RangeColumns>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="StateID" />
</RangeColumns>
<RangeExpressions>
<ScalarOperator ScalarString="@State.[StateID] as [s].[StateID]">
<Identifier>
<ColumnReference Table="@State" Alias="[s]" Column="StateID" />
</Identifier>
</ScalarOperator>
</RangeExpressions>
</Prefix>
</SeekKeys>
</SeekPredicateNew>
</SeekPredicates>
</IndexScan>
</RelOp>
</NestedLoops>
</RelOp>
</Parallelism>
</RelOp>
</Hash>
</RelOp>
</Parallelism>
</RelOp>
</QueryPlan>
</StmtSimple>
以下查询计划适用于估计子树成本为0.14且经过时间为4614毫秒的查询。
<StmtSimple StatementCompId="7" StatementEstRows="339.324" StatementId="4" StatementOptmLevel="FULL" StatementOptmEarlyAbortReason="GoodEnoughPlanFound" StatementSubTreeCost="0.142839" StatementText="SELECT DISTINCT
	t.CountyID,
	t.TownID
FROM dbo.Town t
JOIN @State s
	ON t.StateID = s.StateID
JOIN @County c
	ON t.CountyID = c.CountyID

" StatementType="SELECT" QueryHash="0xF7B13FA8059B141C" QueryPlanHash="0x31CEE9DAF12E77A5">
<StatementSetOptions ANSI_NULLS="true" ANSI_PADDING="true" ANSI_WARNINGS="true" ARITHABORT="true" CONCAT_NULL_YIELDS_NULL="true" NUMERIC_ROUNDABORT="false" QUOTED_IDENTIFIER="true" />
<QueryPlan DegreeOfParallelism="1" MemoryGrant="1584" CachedPlanSize="40" CompileTime="5" CompileCPU="5" CompileMemory="376">
<RelOp AvgRowSize="15" EstimateCPU="0.00454996" EstimateIO="0.0112613" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="339.324" LogicalOp="Distinct Sort" NodeId="0" Parallel="false" PhysicalOp="Sort" EstimatedTotalSubtreeCost="0.142839">
<OutputList>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</OutputList>
<MemoryFractions Input="1" Output="1" />
<RunTimeInformation>
<RunTimeCountersPerThread Thread="0" ActualRebinds="1" ActualRewinds="0" ActualRows="13027" ActualEndOfScans="1" ActualExecutions="1" />
</RunTimeInformation>
<Sort Distinct="true">
<OrderBy>
<OrderByColumn Ascending="true">
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
</OrderByColumn>
<OrderByColumn Ascending="true">
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</OrderByColumn>
</OrderBy>
<RelOp AvgRowSize="15" EstimateCPU="0.0571741" EstimateIO="0" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="339.324" LogicalOp="Inner Join" NodeId="1" Parallel="false" PhysicalOp="Hash Match" EstimatedTotalSubtreeCost="0.127028">
<OutputList>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</OutputList>
<MemoryFractions Input="0" Output="0" />
<RunTimeInformation>
<RunTimeCountersPerThread Thread="0" ActualRows="2110519" ActualEndOfScans="1" ActualExecutions="1" />
</RunTimeInformation>
<Hash>
<DefinedTowns />
<HashKeysBuild>
<ColumnReference Table="@State" Alias="[s]" Column="StateID" />
</HashKeysBuild>
<HashKeysProbe>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="StateID" />
</HashKeysProbe>
<RelOp AvgRowSize="11" EstimateCPU="0.0001581" EstimateIO="0.003125" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="1" LogicalOp="Clustered Index Scan" NodeId="2" Parallel="false" PhysicalOp="Clustered Index Scan" EstimatedTotalSubtreeCost="0.0032831" TableCardinality="0">
<OutputList>
<ColumnReference Table="@State" Alias="[s]" Column="StateID" />
</OutputList>
<RunTimeInformation>
<RunTimeCountersPerThread Thread="0" ActualRows="1" ActualEndOfScans="1" ActualExecutions="1" />
</RunTimeInformation>
<IndexScan Ordered="false" ForcedIndex="false" NoExpandHint="false">
<DefinedTowns>
<DefinedTown>
<ColumnReference Table="@State" Alias="[s]" Column="StateID" />
</DefinedTown>
</DefinedTowns>
<Object Table="[@State]" Index="[PK__#3AB788A__C3BA3B5A3C9FD11A]" Alias="[s]" />
</IndexScan>
</RelOp>
<RelOp AvgRowSize="19" EstimateCPU="0.0357739" EstimateIO="0" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="8558.34" LogicalOp="Inner Join" NodeId="3" Parallel="false" PhysicalOp="Nested Loops" EstimatedTotalSubtreeCost="0.066568">
<OutputList>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="StateID" />
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</OutputList>
<RunTimeInformation>
<RunTimeCountersPerThread Thread="0" ActualRows="2110519" ActualEndOfScans="1" ActualExecutions="1" />
</RunTimeInformation>
<NestedLoops Optimized="false">
<OuterReferences>
<ColumnReference Table="@County" Alias="[c]" Column="CountyID" />
</OuterReferences>
<RelOp AvgRowSize="11" EstimateCPU="0.0001581" EstimateIO="0.003125" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="1" LogicalOp="Clustered Index Scan" NodeId="4" Parallel="false" PhysicalOp="Clustered Index Scan" EstimatedTotalSubtreeCost="0.0032831" TableCardinality="0">
<OutputList>
<ColumnReference Table="@County" Alias="[c]" Column="CountyID" />
</OutputList>
<RunTimeInformation>
<RunTimeCountersPerThread Thread="0" ActualRows="137" ActualEndOfScans="1" ActualExecutions="1" />
</RunTimeInformation>
<IndexScan Ordered="false" ForcedIndex="false" NoExpandHint="false">
<DefinedTowns>
<DefinedTown>
<ColumnReference Table="@County" Alias="[c]" Column="CountyID" />
</DefinedTown>
</DefinedTowns>
<Object Table="[@County]" Index="[PK__#3E88198__B68F9DF7407061FE]" Alias="[c]" />
</IndexScan>
</RelOp>
<RelOp AvgRowSize="19" EstimateCPU="0.00957118" EstimateIO="0.0179398" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="8558.34" LogicalOp="Index Seek" NodeId="5" Parallel="false" PhysicalOp="Index Seek" EstimatedTotalSubtreeCost="0.027511" TableCardinality="6291320">
<OutputList>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="StateID" />
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</OutputList>
<RunTimeInformation>
<RunTimeCountersPerThread Thread="0" ActualRows="2110519" ActualEndOfScans="137" ActualExecutions="137" />
</RunTimeInformation>
<IndexScan Ordered="true" ScanDirection="FORWARD" ForcedIndex="false" ForceSeek="false" NoExpandHint="false">
<DefinedTowns>
<DefinedTown>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="StateID" />
</DefinedTown>
<DefinedTown>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
</DefinedTown>
<DefinedTown>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="TownID" />
</DefinedTown>
</DefinedTowns>
<Object Database="[Locality]" Schema="[dbo]" Table="[Town]" Index="[NCX_Town_CountyID_inc_StateID_TownID]" Alias="[t]" IndexKind="NonClustered" />
<SeekPredicates>
<SeekPredicateNew>
<SeekKeys>
<Prefix ScanType="EQ">
<RangeColumns>
<ColumnReference Database="[Locality]" Schema="[dbo]" Table="[Town]" Alias="[t]" Column="CountyID" />
</RangeColumns>
<RangeExpressions>
<ScalarOperator ScalarString="@County.[CountyID] as [c].[CountyID]">
<Identifier>
<ColumnReference Table="@County" Alias="[c]" Column="CountyID" />
</Identifier>
</ScalarOperator>
</RangeExpressions>
</Prefix>
</SeekKeys>
</SeekPredicateNew>
</SeekPredicates>
</IndexScan>
</RelOp>
</NestedLoops>
</RelOp>
</Hash>
</RelOp>
</Sort>
</RelOp>
</QueryPlan>
</StmtSimple>
答案 0 :(得分:6)
虽然我没有深入研究您的查询计划,但我有建议和建议的解释。建议很简单:不要将表变量用于行数超过两个数量级(100行)的表。可能的解释是,它们的性能可能非常差,并且生成的查询计划可能不准确或效率低下,per Microsoft's article on table data types:
最佳做法
不要使用表变量来存储大量数据(超过100行)。当表变量包含大量数据时,计划选择可能不是最佳的或稳定的。考虑重写此类查询以使用临时表或使用USE PLAN查询提示来确保优化程序使用适用于您的方案的现有查询计划。
和
限制和限制
SQL Server优化器基于成本的推理模型不支持表变量。因此,当需要基于成本的选择来实现有效的查询计划时,不应使用它们。当需要基于成本的选择时,首选临时表。这通常包括具有连接,并行性决策和索引选择选择的查询。
修改表变量的查询不会生成并行查询执行计划。当修改非常大的表变量或复杂查询中的表变量时,性能会受到影响。在这些情况下,请考虑使用临时表。有关更多信息,请参见CREATE TABLE(Transact-SQL)。读取表变量而不修改它们的查询仍然可以并行化。
...
实际上,我发现使用表变量会排除生成并行查询计划,并且会损害除最简单,最不复杂的表变量之外的所有变量。
相反,我发现临时表几乎在每种情况下都更高效,并且只要您小心地及时清理自己(或使用短期会话以允许服务器清除临时表) )然后你会获得更好的表现。
至于为什么第二个查询计划的成本更低但执行时间更长?可能有很多很多县(1000s?10000s?),生成的查询计划假设有一个很小的数字(100s)。查询优化器认为 @County
可能少于100行,因此可能认为结果集非常具体,因为列表城镇加入了一个&lt; 100条目的列表,然后加入第二个&lt; 100条目列表,很可能有少于100行。
实际上,每个城镇都有一个县,因此连接不是很有选择性,并且使用连接类型的不良选择。在每种情况下,嵌套循环用于最后一次连接,但嵌套循环连接对于大型排序数据效率极低。这可能是问题的症结所在。
使用临时表而不是表变量重新运行您的查询,我猜您将获得具有非常不同的性能特征的非常不同的查询计划。
答案 1 :(得分:-3)
Per Mikael的评论。第一个查询更快,但成本更高,因为它使用并行性。