我有一个长期运行的SQL Server 2005查询,我一直希望优化它。
当我查看实际的执行计划时,它表示Clustered Index Seek有66%的成本。
执行计划Snippit:
<RelOp AvgRowSize="31" EstimateCPU="0.0113754" EstimateIO="0.0609028" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="10198.5" LogicalOp="Clustered Index Seek" NodeId="16" Parallel="false" PhysicalOp="Clustered Index Seek" EstimatedTotalSubtreeCost="0.0722782">
<OutputList>
<ColumnReference Database="[wf_1]" Schema="[dbo]" Table="[Indices]" Alias="[I]" Column="quoteDate" />
<ColumnReference Database="[wf_1]" Schema="[dbo]" Table="[Indices]" Alias="[I]" Column="price" />
<ColumnReference Database="[wf_1]" Schema="[dbo]" Table="[Indices]" Alias="[I]" Column="tenure" />
</OutputList>
<RunTimeInformation>
<RunTimeCountersPerThread Thread="0" ActualRows="1067" ActualEndOfScans="1" ActualExecutions="1" />
</RunTimeInformation>
<IndexScan Ordered="true" ScanDirection="FORWARD" ForcedIndex="false" NoExpandHint="false">
<DefinedValues>
<DefinedValue>
<ColumnReference Database="[wf_1]" Schema="[dbo]" Table="[Indices]" Alias="[I]" Column="quoteDate" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[wf_1]" Schema="[dbo]" Table="[Indices]" Alias="[I]" Column="price" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[wf_1]" Schema="[dbo]" Table="[Indices]" Alias="[I]" Column="tenure" />
</DefinedValue>
</DefinedValues>
<Object Database="[wf_1]" Schema="[dbo]" Table="[Indices]" Index="[_dta_index_Indices_14_320720195__K5_K2_K1_3]" Alias="[I]" />
<SeekPredicates>
<SeekPredicate>
<Prefix ScanType="EQ">
<RangeColumns>
<ColumnReference Database="[wf_1]" Schema="[dbo]" Table="[Indices]" Alias="[I]" Column="HedgeProduct" ComputedColumn="true" />
</RangeColumns>
<RangeExpressions>
<ScalarOperator ScalarString="(1)">
<Const ConstValue="(1)" />
</ScalarOperator>
</RangeExpressions>
</Prefix>
<StartRange ScanType="GE">
<RangeColumns>
<ColumnReference Database="[wf_1]" Schema="[dbo]" Table="[Indices]" Alias="[I]" Column="tenure" />
</RangeColumns>
<RangeExpressions>
<ScalarOperator ScalarString="[@StartMonth]">
<Identifier>
<ColumnReference Column="@StartMonth" />
</Identifier>
</ScalarOperator>
</RangeExpressions>
</StartRange>
<EndRange ScanType="LE">
<RangeColumns>
<ColumnReference Database="[wf_1]" Schema="[dbo]" Table="[Indices]" Alias="[I]" Column="tenure" />
</RangeColumns>
<RangeExpressions>
<ScalarOperator ScalarString="[@EndMonth]">
<Identifier>
<ColumnReference Column="@EndMonth" />
</Identifier>
</ScalarOperator>
</RangeExpressions>
</EndRange>
</SeekPredicate>
</SeekPredicates>
</IndexScan>
</RelOp>
从这个,有没有人看到一个明显的问题,导致这需要这么长时间?
以下是查询:
(SELECT quotedate, tenure, price, ActualVolume, HedgePortfolioValue, Price AS UnhedgedPrice, ((ActualVolume*Price - HedgePortfolioValue)/ActualVolume) AS HedgedPrice
FROM
(
SELECT [quoteDate]
,[price]
, tenure
,isnull(wf_1.[Risks].[HedgePortValueAsOfDate2](1,tenureMonth,quotedate,price),0) as HedgePortfolioValue
,[TotalOperatingGasVolume] as ActualVolume
FROM [wf_1].[dbo].[Indices] I
inner join
(
SELECT DISTINCT tenureMonth
FROM [wf_1].[Risks].[KnowRiskTrades]
WHERE HedgeProduct = 1
AND portfolio <> 'Natural Gas Hedge Transactions'
) B ON I.tenure=B.tenureMonth
inner join
(
SELECT [Month],[TotalOperatingGasVolume]
FROM [wf_1].[Risks].[ActualGasVolumes]
) C ON C.[Month]=B.tenureMonth
WHERE HedgeProduct = 1
AND quoteDate>=dateadd(day, -3*365, tenureMonth)
AND quoteDate<=dateadd(day,-3,tenureMonth)
)A
)
答案 0 :(得分:1)
听起来您的统计信息已关闭。
该查询正在扼杀我的眼睛,我不想猜测并选择错误的表格,无论CI寻求的是哪个表格,都至少更新该表格的统计数据并再次检查查询。
更新统计数据: http://msdn.microsoft.com/en-us/library/ms187348.aspx
答案 1 :(得分:1)
这一点引起了人们的注意,你有投资组合的指数吗?你为什么使用DISTINCT?
SELECT DISTINCT tenureMonth
FROM [wf_1].[Risks].[KnowRiskTrades]
WHERE HedgeProduct = 1
AND portfolio <> 'Natural Gas Hedge Transactions'
这样做:
WHERE HedgeProduct = 1
AND quoteDate>=dateadd(day, -3*365, tenureMonth)
AND quoteDate<=dateadd(day,-3,tenureMonth)
但是,如果不了解索引的位置以及相关表中的数据量,很难提供好的建议。
另外:查询需要多长时间?你想要多长时间?你多久运行一次?您的数据库服务器有多忙?