那些遇到过的问题

为什么 ISNULL 会提高查询的性能？

F.B*_*ate 5 t-sql performance sql-server-2005

我们目前正在尝试加速一些查询，但我们遇到了一些我（不是 DBA 只是 .NET 开发人员）无法解释或理解的问题。我们在 SQL Server 2005 上运行这个查询。

我们有以下查询（为了论证而变得小而简单）；

SELECT    
    *
FROM    
    RandomTable 
WHERE   
    MoneyColumn <> 0
GROUP BY
    SomeColumn
Run Code Online (Sandbox Code Playgroud)

这个查询在大约三秒钟内运行，然后我们随机尝试跟随以加快速度（真的是在黑暗中拍摄）

SELECT    
    *
FROM    
    RandomTable 
WHERE   
    isnull(MoneyColumn,0) <> 0
GROUP BY
    SomeColumn
Run Code Online (Sandbox Code Playgroud)

这降低了查询速度围绕的一个第二个..

该列没有 NULL 值（但由于数据库设计非常糟糕）它却是 NULLABLE ...

事实上，它是 NULLABLE 使 SQL Server 做一些事情来解决这个问题，使它在没有提到 ISNULL 的情况下变慢吗？我只是不知道为什么 ISNULL 会使它执行得更快（而且幅度如此之大）。我认为当查询中有 ISNULL 语句时，SQL 实际上有更多的工作要做。

任何人都可以对此有所了解吗？

添加了编辑执行计划

带 ISNULL

<?xml version="1.0" encoding="utf-16"?>
<ShowPlanXML xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema" Version="1.0" Build="9.00.5000.00" xmlns="http://schemas.microsoft.com/sqlserver/2004/07/showplan">
  <BatchSequence>
    <Batch>
      <Statements>
        <StmtSimple StatementCompId="1" StatementEstRows="9019.76" StatementId="1" StatementOptmLevel="FULL" StatementSubTreeCost="1.48105" StatementText="SELECT    debiteur_id, MIN(Faktuurdatum) AS OldestOpenInvoiceDate, ISNULL(SUM(Totaal_Open),0) AS TotalOpenAmount&#xD;&#xA;FROM    dbo.tbl_Faktuur &#xD;&#xA;WHERE   (Afgehandeld_NeeJa = 0 OR Afgehandeld_NeeJa IS NULL)&#xD;&#xA;AND        (ISNULL(Totaal_Open,0) &lt;&gt; 0) &#xD;&#xA;--AND        (Totaal_Open &lt;&gt; 0) &#xD;&#xA;GROUP BY debiteur_id" StatementType="SELECT">
          <StatementSetOptions ANSI_NULLS="false" ANSI_PADDING="false" ANSI_WARNINGS="false" ARITHABORT="true" CONCAT_NULL_YIELDS_NULL="false" NUMERIC_ROUNDABORT="false" QUOTED_IDENTIFIER="false" />
          <QueryPlan DegreeOfParallelism="1" MemoryGrant="1520" CachedPlanSize="54" CompileTime="11" CompileCPU="11" CompileMemory="704">
            <RelOp AvgRowSize="23" EstimateCPU="0.000901976" EstimateIO="0" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="9019.76" LogicalOp="Compute Scalar" NodeId="0" Parallel="false" PhysicalOp="Compute Scalar" EstimatedTotalSubtreeCost="1.48105">
              <OutputList>
                <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Debiteur_ID" />
                <ColumnReference Column="Expr1003" />
                <ColumnReference Column="Expr1005" />
              </OutputList>
              <ComputeScalar>
                <DefinedValues>
                  <DefinedValue>
                    <ColumnReference Column="Expr1005" />
                    <ScalarOperator ScalarString="isnull([Expr1004],($0.0000))">
                      <Intrinsic FunctionName="isnull">
                        <ScalarOperator>
                          <Identifier>
                            <ColumnReference Column="Expr1004" />
                          </Identifier>
                        </ScalarOperator>
                        <ScalarOperator>
                          <Const ConstValue="($0.0000)" />
                        </ScalarOperator>
                      </Intrinsic>
                    </ScalarOperator>
                  </DefinedValue>
                </DefinedValues>
                <RelOp AvgRowSize="23" EstimateCPU="0.291662" EstimateIO="0" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="9019.76" LogicalOp="Compute Scalar" NodeId="1" Parallel="false" PhysicalOp="Compute Scalar" EstimatedTotalSubtreeCost="1.48014">
                  <OutputList>
                    <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Debiteur_ID" />
                    <ColumnReference Column="Expr1003" />
                    <ColumnReference Column="Expr1004" />
                  </OutputList>
                  <ComputeScalar>
                    <DefinedValues>
                      <DefinedValue>
                        <ColumnReference Column="Expr1004" />
                        <ScalarOperator ScalarString="CASE WHEN [Expr1013]=(0) THEN NULL ELSE [Expr1014] END">
                          <IF>
                            <Condition>
                              <ScalarOperator>
                                <Compare CompareOp="EQ">
                                  <ScalarOperator>
                                    <Identifier>
                                      <ColumnReference Column="Expr1013" />
                                    </Identifier>
                                  </ScalarOperator>
                                  <ScalarOperator>
                                    <Const ConstValue="(0)" />
                                  </ScalarOperator>
                                </Compare>
                              </ScalarOperator>
                            </Condition>
                            <Then>
                              <ScalarOperator>
                                <Const ConstValue="NULL" />
                              </ScalarOperator>
                            </Then>
                            <Else>
                              <ScalarOperator>
                                <Identifier>
                                  <ColumnReference Column="Expr1014" />
                                </Identifier>
                              </ScalarOperator>
                            </Else>
                          </IF>
                        </ScalarOperator>
                      </DefinedValue>
                    </DefinedValues>
                    <RelOp AvgRowSize="23" EstimateCPU="0.291662" EstimateIO="0" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="9019.76" LogicalOp="Aggregate" NodeId="2" Parallel="false" PhysicalOp="Hash Match" EstimatedTotalSubtreeCost="1.48014">
                      <OutputList>
                        <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Debiteur_ID" />
                        <ColumnReference Column="Expr1003" />
                        <ColumnReference Column="Expr1013" />
                        <ColumnReference Column="Expr1014" />
                      </OutputList>
                      <MemoryFractions Input="1" Output="1" />
                      <RunTimeInformation>
                        <RunTimeCountersPerThread Thread="0" ActualRows="156794" ActualEndOfScans="1" ActualExecutions="1" />
                      </RunTimeInformation>
                      <Hash>
                        <DefinedValues>
                          <DefinedValue>
                            <ColumnReference Column="Expr1003" />
                            <ScalarOperator ScalarString="MIN([directpay].[dbo].[tbl_Faktuur].[Faktuurdatum])">
                              <Aggregate AggType="MIN" Distinct="false">
                                <ScalarOperator>
                                  <Identifier>
                                    <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Faktuurdatum" />
                                  </Identifier>
                                </ScalarOperator>
                              </Aggregate>
                            </ScalarOperator>
                          </DefinedValue>
                          <DefinedValue>
                            <ColumnReference Column="Expr1013" />
                            <ScalarOperator ScalarString="COUNT_BIG([directpay].[dbo].[tbl_Faktuur].[Totaal_Open])">
                              <Aggregate AggType="COUNT_BIG" Distinct="false">
                                <ScalarOperator>
                                  <Identifier>
                                    <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Totaal_Open" />
                                  </Identifier>
                                </ScalarOperator>
                              </Aggregate>
                            </ScalarOperator>
                          </DefinedValue>
                          <DefinedValue>
                            <ColumnReference Column="Expr1014" />
                            <ScalarOperator ScalarString="SUM([directpay].[dbo].[tbl_Faktuur].[Totaal_Open])">
                              <Aggregate AggType="SUM" Distinct="false">
                                <ScalarOperator>
                                  <Identifier>
                                    <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Totaal_Open" />
                                  </Identifier>
                                </ScalarOperator>
                              </Aggregate>
                            </ScalarOperator>
                          </DefinedValue>
                        </DefinedValues>
                        <HashKeysBuild>
                          <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Debiteur_ID" />
                        </HashKeysBuild>
                        <RelOp AvgRowSize="23" EstimateCPU="0.255" EstimateIO="0.634196" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="27420" LogicalOp="Index Seek" NodeId="4" Parallel="false" PhysicalOp="Index Seek" EstimatedTotalSubtreeCost="0.889196">
                          <OutputList>
                            <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Debiteur_ID" />
                            <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Faktuurdatum" />
                            <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Totaal_Open" />
                          </OutputList>
                          <RunTimeInformation>
                            <RunTimeCountersPerThread Thread="0" ActualRows="298726" ActualEndOfScans="1" ActualExecutions="1" />
                          </RunTimeInformation>
                          <IndexScan Ordered="true" ScanDirection="FORWARD" ForcedIndex="false" NoExpandHint="false">
                            <DefinedValues>
                              <DefinedValue>
                                <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Debiteur_ID" />
                              </DefinedValue>
                              <DefinedValue>
                                <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Faktuurdatum" />
                              </DefinedValue>
                              <DefinedValue>
                                <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Totaal_Open" />
                              </DefinedValue>
                            </DefinedValues>
                            <Object Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Index="[_dta_index_tbl_Faktuur_5_583009158__K13_K9_K19_K2_5]" />
                            <SeekPredicates>
                              <SeekPredicate>
                                <Prefix ScanType="EQ">
                                  <RangeColumns>
                                    <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Afgehandeld_NeeJa" />
                                  </RangeColumns>
                                  <RangeExpressions>
                                    <ScalarOperator ScalarString="(0)">
                                      <Const ConstValue="(0)" />
                                    </ScalarOperator>
                                  </RangeExpressions>
                                </Prefix>
                              </SeekPredicate>
                            </SeekPredicates>
                            <Predicate>
                              <ScalarOperator ScalarString="isnull([directpay].[dbo].[tbl_Faktuur].[Totaal_Open],($0.0000))&lt;($0.0000) OR isnull([directpay].[dbo].[tbl_Faktuur].[Totaal_Open],($0.0000))&gt;($0.0000)">
                                <Logical Operation="OR">
                                  <ScalarOperator>
                                    <Compare CompareOp="LT">
                                      <ScalarOperator>
                                        <Intrinsic FunctionName="isnull">
                                          <ScalarOperator>
                                            <Identifier>
                                              <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Totaal_Open" />
                                            </Identifier>
                                          </ScalarOperator>
                                          <ScalarOperator>
                                            <Const ConstValue="($0.0000)" />
                                          </ScalarOperator>
                                        </Intrinsic>
                                      </ScalarOperator>
                                      <ScalarOperator>
                                        <Const ConstValue="($0.0000)" />
                                      </ScalarOperator>
                                    </Compare>
                                  </ScalarOperator>
                                  <ScalarOperator>
                                    <Compare CompareOp="GT">
                                      <ScalarOperator>
                                        <Intrinsic FunctionName="isnull">
                                          <ScalarOperator>
                                            <Identifier>
                                              <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Totaal_Open" />
                                            </Identifier>
                                          </ScalarOperator>
                                          <ScalarOperator>
                                            <Const ConstValue="($0.0000)" />
                                          </ScalarOperator>
                                        </Intrinsic>
                                      </ScalarOperator>
                                      <ScalarOperator>
                                        <Const ConstValue="($0.0000)" />
                                      </ScalarOperator>
                                    </Compare>
                                  </ScalarOperator>
                                </Logical>
                              </ScalarOperator>
                            </Predicate>
                          </IndexScan>
                        </RelOp>
                      </Hash>
                    </RelOp>
                  </ComputeScalar>
                </RelOp>
              </ComputeScalar>
            </RelOp>
          </QueryPlan>
        </StmtSimple>
      </Statements>
    </Batch>
  </BatchSequence>
</ShowPlanXML>
Run Code Online (Sandbox Code Playgroud)

没有 ISNULL

<?xml version="1.0" encoding="utf-16"?>
<ShowPlanXML xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema" Version="1.0" Build="9.00.5000.00" xmlns="http://schemas.microsoft.com/sqlserver/2004/07/showplan">
  <BatchSequence>
    <Batch>
      <Statements>
        <StmtSimple StatementCompId="1" StatementEstRows="1322.43" StatementId="1" StatementOptmLevel="FULL" StatementOptmEarlyAbortReason="GoodEnoughPlanFound" StatementSubTreeCost="0.274954" StatementText="SELECT    debiteur_id, MIN(Faktuurdatum) AS OldestOpenInvoiceDate, ISNULL(SUM(Totaal_Open),0) AS TotalOpenAmount&#xD;&#xA;FROM    dbo.tbl_Faktuur &#xD;&#xA;WHERE   (Afgehandeld_NeeJa = 0 OR Afgehandeld_NeeJa IS NULL)&#xD;&#xA;--AND        (ISNULL(Totaal_Open,0) &lt;&gt; 0) &#xD;&#xA;AND        (Totaal_Open &lt;&gt; 0) &#xD;&#xA;GROUP BY debiteur_id" StatementType="SELECT">
          <StatementSetOptions ANSI_NULLS="false" ANSI_PADDING="false" ANSI_WARNINGS="false" ARITHABORT="true" CONCAT_NULL_YIELDS_NULL="false" NUMERIC_ROUNDABORT="false" QUOTED_IDENTIFIER="false" />
          <QueryPlan CachedPlanSize="47" CompileTime="9" CompileCPU="9" CompileMemory="528">
            <RelOp AvgRowSize="23" EstimateCPU="0.000132243" EstimateIO="0" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="1322.43" LogicalOp="Compute Scalar" NodeId="0" Parallel="false" PhysicalOp="Compute Scalar" EstimatedTotalSubtreeCost="0.274954">
              <OutputList>
                <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Debiteur_ID" />
                <ColumnReference Column="Expr1003" />
                <ColumnReference Column="Expr1005" />
              </OutputList>
              <ComputeScalar>
                <DefinedValues>
                  <DefinedValue>
                    <ColumnReference Column="Expr1005" />
                    <ScalarOperator ScalarString="isnull([Expr1004],($0.0000))">
                      <Intrinsic FunctionName="isnull">
                        <ScalarOperator>
                          <Identifier>
                            <ColumnReference Column="Expr1004" />
                          </Identifier>
                        </ScalarOperator>
                        <ScalarOperator>
                          <Const ConstValue="($0.0000)" />
                        </ScalarOperator>
                      </Intrinsic>
                    </ScalarOperator>
                  </DefinedValue>
                </DefinedValues>
                <RelOp AvgRowSize="23" EstimateCPU="0.167304" EstimateIO="0" EstimateRebinds="0" EstimateRewinds="0" EstimateRows="1322.43" LogicalOp="Aggregate" NodeId="1" Parallel="false" PhysicalOp="Hash Match" EstimatedTotalSubtreeCost="0.274822">
                  <OutputList>
                    <ColumnReference Database="[directpay]" Schema="[dbo]" Table="[tbl_Faktuur]" Column="Debiteur_ID" />
                    <ColumnReference Column="Expr1003" />
                    <ColumnReference Column="Expr1004" />
                  </OutputList>
                  <MemoryFractions In

Mar*_*ith 3

该ISNULL()版本使谓词的该部分成为unsargable。

这也意味着它无法有效地使用索引（它仍然可以从查询中的其他谓词查找前导列），这也使得在估计行数时构造对于优化器来说更加不透明会匹配的。

这两个计划都有估计 27,420 行进入哈希匹配运算符，但它们的不同之处在于输出的估计行数。较快的计划预计 9,019 人，较慢的计划预计 1,322 人。

事实上，这两个估计都是非常错误的，因为实际上按列分组有 156,794 个不同的值。由于错误的估计，两个查询都没有分配足够的内存来构建哈希表，并且都需要溢出到磁盘。速度较慢的程序特别缺乏内存，因此这种情况会发生多次。

您可以尝试更新现有统计信息或查看创建一些多列统计信息，以便尝试更准确地估计将返回的不同分组依据项目的数量，从而获得不同的计划或分配更多内存。

归档时间：

查看次数：

1651 次

最近记录：

12 年，10 月 前
相关归档
哪个在Python中更快:x**.5还是math.sqrt(x)？ 175
从sql server中的字符串中删除html标签的最佳方法？ 107
查询SQL Server上存储过程的内容 29
为什么执行计划包含用于持久保存的计算列的用户定义函数调用？ 18
从历史记录表中选择最新状态 10
如何使用case时从tsql返回true或false 7
英特尔编译器生成的代码比MSVC慢68%(提供完整示例) 7
如何通过存储过程执行数据库中的所有视图 4
如何使用 php、SQL SERVER 2005 从 INSERT 语句返回插入的行？ 2
如何使同一项目的多次出现与第一次出现的值不同？ 2
难疑归档
如何退出Vim编辑器？ 3558
如何制作一系列功能装饰器？ 2647
你如何设置,清除和切换一个位？ 2454
finally块总是在Java中执行吗？ 2281
如何在Java中声明和初始化数组？ 1946
<快于<=？ 1508
编译用于高放射性环境的应用程序 1414
创建将T限制为枚举的通用方法 1122
Access-Control-Allow-Origin标头如何工作？ 1050
如何在JavaScript中将浮点数转换为整数？ 1043