SqlServer性能优化 即席查询（十三）
Posted 石
tags:
篇首语：本文由小常识网(cha138.com)小编为大家整理，主要介绍了SqlServer性能优化 即席查询（十三）相关的知识，希望对你有一定的参考价值。
执行计划，查询类别：
1.即席查询 2.预定义查询
select c.EnglishProductCategoryName,p.EnglishProductName,p.Color,p.Size from Product as p inner join ProductCategory as c on p.ProductSubcategoryKey= c.ProductCategoryKey where p.Size>\'1\'
--查询执行计划是否被缓存 select c.usecounts,c.size_in_bytes,c.objtype,t.text from sys.dm_exec_cached_plans as c cross apply sys.dm_exec_sql_text(c.plan_handle) as t dbcc freeproccache--清空执行计划
--没有join 的形式会生成简单参数化 select EnglishProductName,Color,Size from Product where size>\'1\'--简单参数化
select EnglishProductName,Color,Size from Product where size>\'2\'--简单参数化
select c.EnglishProductCategoryName,p.EnglishProductName,p.Color,p.Size from Product as p inner join ProductCategory as c on p.aProductSubcategoryKey= c.ProductCategoryKey where p.Size>\'2\'
语句一样时即席查询才会重用执行计划。
优化：打开开关
exec sp_configure \'show advanced options\',1 reconfigure with override
为ad hoc的查询优化：
exec sp_configure \'Optimize for ad hoc workloads\',1 reconfigure with override
--使用参数化 alter database HRDB set Parameterization forced
set Parameterization forced 强制参数化（like无法识别 ）
select c.EnglishProductCategoryName,p.EnglishProductName,p.Color,p.Size from Product as p inner join ProductCategory as c on p.ProductSubcategoryKey= c.ProductCategoryKey where p.Size>\'2\' select c.EnglishProductCategoryName,p.EnglishProductName,p.Color,p.Size from Product as p inner join ProductCategory as c on p.ProductSubcategoryKey= c.ProductCategoryKey where p.Size like \'2%\'
预定义查询：
预定义查询--参数化执行计划：
存储过程：
1.创建时延时检查
2.第一次执行时编译并生成执行计划
3.减少网络传输量
4.封装变化点
5.增强安全性，隔离访问控制
创建存储过程：
create procedure p_querycp @size varchar(500) as select c.EnglishProductCategoryName,p.EnglishProductName,p.Color,p.size from Product as p inner join ProductCategory as c on p.ProductSubcategoryKey=c.ProductCategoryKey where p.Size>@size
做跟踪(以前有对应得截图)：
执行存储过程：
create procedure p_querycp @size varchar(500) as select c.EnglishProductCategoryName,p.EnglishProductName,p.Color,p.size from Product as p inner join ProductCategory as c on p.ProductSubcategoryKey=c.ProductCategoryKey where p.Size>@size --清空执行计划 dbcc freeproccache --执行 exec p_querycp \'1\'
执行重复的语句：
dbcc freeproccache exec p_querycp @size=\'1\' exec p_querycp @size=\'2\'
查看缓存计划：
select c.usecounts,c.size_in_bytes,c.objtype,t.text from sys.dm_exec_cached_plans as c cross apply sys.dm_exec_sql_text(c.plan_handle) as t
预定义查询---参数化执行计划：
SP_ExecuteSql
避免了自己维护存储过程管理成本
可重用执行计划
Unicode字符串作为参数值与类型
大小写敏感
把存储过程定义成传递参数的：
declare @sqltext nvarchar(500) set @sqltext=N\' select c.EnglishProductCategoryName,p.EnglishProductName,p.Color,p.size from Product as p inner join ProductCategory as c on p.ProductSubcategoryKey=c.ProductCategoryKey where p.Size>@size \' declare @params nvarchar(500) set @params=N\'@size varchar(500)\' exec sp_executesql @sqltext,@params,@size=\'1\'
把size 的大小换成 2
在.net中调用：（两种写法）
public object getCp(string size)
{
HRUser dbcontext = new HRUser();
var cps = from p in dbcontext.Product
join c in dbcontext.ProductCategory
on p.ProductSubcategoryKey equals c.ProductCategoryKey
where p.Size == size
//返回匿名对象
select new
{
CName = c.EnglishProductCategoryName,
PName = p.EnglishProductName,
Color = p.Color,
Size = p.Size
};
return cps.ToList();
}
public object getcp(string size)
{
HRUser dbcontext = new HRUser();
var cps = dbcontext.Product.Join(dbcontext.ProductCategory, a => a.ProductSubcategoryKey, ar => ar.ProductCategoryKey, (a, ar) => new
{
CName = ar.EnglishProductCategoryName,
PName = a.EnglishProductName,
Color = a.Color,
Size = a.Size
}).Where(p => p.Size == size);
return cps.ToList();
}
页面：
<asp:TextBox ID="TextBox1" runat="server"></asp:TextBox>
<asp:Button ID="Button2" runat="server" OnClick="Button2_Click" Text="显示产品" />
<asp:GridView ID="GridView1" runat="server">
</asp:GridView>
点击后的事件：
protected void Button2_Click(object sender, EventArgs e)
{
Product p = new Product();
var cps = p.getCp(TextBox1.Text.Trim());
GridView1.DataSource = cps;
GridView1.DataBind();
}
--动态构建语句（执行带参数的方法） declare @size varchar(500) set @size=\'2\' execute(\'select c.EnglishProductCategoryName,p.EnglishProductName,p.Color,p.size from Product as p inner join ProductCategory as c on p.ProductSubcategoryKey=c.ProductCategoryKey where p.Size>\'+\'\'\'\'+@size+\'\'\'\')
dbcc freeproccache --执行计划 缓存 select c.usecounts,c.size_in_bytes,c.objtype,t.text from sys.dm_exec_cached_plans as c cross apply sys.dm_exec_sql_text(c.plan_handle) as t
形成两个缓存计划：
创建存储过程：
create procedure p_querye @vacationhours int as select e.LoginID,e.JobTitle from EmployeeOp as e where VacationHours>@vacationhours
exec p_querye 2--实际执行计划 表扫描 exec p_querye 99--实际执行计划 表扫描 应用用索引更好
--重新编译的执行计划 exec p_querycp 99 with recompile
手工的指定执行几乎：
--手工的指定执行计划 exec sp_create_plan_guide @name=\'执行任务计划指南之EmployeeOp Vacation\', --转成Unicode编码格式 @stmt=N\'select e.LoginID,e.JobTitle from EmployeeOp as e where VacationHours>@vacationhours\', @type=N\'Object\', --执行计划的名字 @module_or_batch =\'p_querye\', @params =null, --提示 @hints =N\'OPTION(OPTIMIZE FOR(@vacationhours=\'\'99\'\'))\'
清除执行计划：执行（会参考上面指定的执行计划）
exec p_querye 2
存储过程重编译：
临时结果集：
定义跟踪的模板：
一：使用物理表进行临时结果集
--1.做一张物理表 create procedure p_physicaltb as CREATE TABLE PhysicalTB( [SalesOrderID] [int] NOT NULL, [SalesOrderDetailID] [int] NOT NULL, [CarrierTrackingNumber] [nvarchar](25) NULL, [OrderQty] [smallint] NOT NULL, [ProductID] [int] NOT NULL, [SpecialOfferID] [int] NOT NULL, [UnitPrice] [money] NOT NULL, [UnitPriceDiscount] [money] NOT NULL, [LineTotal] [numeric](38, 6) NOT NULL, [rowguid] [uniqueidentifier] NOT NULL, [ModifiedDate] [datetime] NOT NULL ) insert into PhysicalTB select * from OrderDetail select * from PhysicalTB set statistics time on exec p_physicaltb --cpu:391 total:1762 set statistics time off
跟踪的情况：
删除之后再次创建执行。
物理表每次执行都会有重编译的过程（不建议使用物理表来存储临时结果集）
第二种方式：
临时表存储临时结果集
create procedure p_temptb as CREATE TABLE #PhysicalTB( [SalesOrderID] [int] NOT NULL, [SalesOrderDetailID] [int] NOT NULL, [CarrierTrackingNumber] [nvarchar](25) NULL, [OrderQty] [smallint] NOT NULL, [ProductID] [int] NOT NULL, [SpecialOfferID] [int] NOT NULL, [UnitPrice] [money] NOT NULL, [UnitPriceDiscount] [money] NOT NULL, [LineTotal] [numeric](38, 6) NOT NULL, [rowguid] [uniqueidentifier] NOT NULL, [ModifiedDate] [datetime] NOT NULL ) insert into #PhysicalTB select * from OrderDetail select * from #PhysicalTB drop table #PhysicalTB set statistics time on exec p_temptb --cpu:110 total:1494 set statistics time off sp_helpdb \'tempdb\'--16064,768
第一次执行时重新编译，第二次就不会重新编译了。
用到了tempdb临时表：
第三种方式：表变量存储临时结果集
--表变量存储临时结果集 create procedure p_tabletb as --申明表变量 declare @PhysicalTB table( [SalesOrderID] [int] NOT NULL, [SalesOrderDetailID] [int] NOT NULL, [CarrierTrackingNumber] [nvarchar](25) NULL, [OrderQty] [smallint] NOT NULL, [ProductID] [int] NOT NULL, [SpecialOfferID] [int] NOT NULL, [UnitPrice] [money] NOT NULL, [UnitPriceDiscount] [money] NOT NULL, [LineTotal] [numeric](38, 6) NOT NULL, [rowguid] [uniqueidentifier] NOT NULL, [ModifiedDate] [datetime] NOT NULL ) insert into @PhysicalTB select * from OrderDetail select * from @PhysicalTB set statistics time on exec p_tabletb --cpu:110 total:1494 set statistics time off sp_helpdb \'tempdb\'--17064,768
执行时不会重新编译
第四种方式：
--CTE(通用表表达式)存储临时结果集 完全放在内存中 不会操作任何数据库中的东西 create procedure p_ctetb as begin --会自动推断数据类型 ;with PhysicalTB( [SalesOrderID], [SalesOrderDetailID], [CarrierTrackingNumber], [OrderQty] , [ProductID], [SpecialOfferID], [UnitPrice] , [UnitPriceDiscount] , [LineTotal] , [rowguid] , [ModifiedDate] ) as (select * from OrderDetail) --访问通用表表达式 select * from PhysicalTB end
跟踪的结果：
没有重新编译的过程，纯粹操作内存。tempdb数据库不会有任何的变化。
set statistics time on exec p_ctetb --cpu:100 total:1300 set statistics time off sp_helpdb \'tempdb\'--17064,768
高版本的通用表达式可以进行多次的使用：
create procedure p_ctetb1 as begin ;with PhysicalTB( [SalesOrderID], [SalesOrderDetailID], [CarrierTrackingNumber], [OrderQty] , [ProductID], [SpecialOfferID], [UnitPrice] , [UnitPriceDiscount] , [LineTotal] , [rowguid] , [ModifiedDate] ) as (select * from OrderDetail) select * from PhysicalTB select * from PhysicalTB end exec p_ctetb1
08之前的数据库，只要把表达式在创建一次就可以了
临时数据集的优化处理：
优化查询：编译指南。
--清空执行计划
dbcc freeproccache
select * from EmployeeOp where VacationHours>1 option(use plan N\'
<ShowPlanXML xmlns="http://schemas.microsoft.com/sqlserver/2004/07/showplan" Version="1.2" Build="12.0.2000.8">
<BatchSequence>
<Batch>
<Statements>
<StmtSimple StatementText="select * from EmployeeOp where VacationHours>99" StatementId="1" StatementCompId="1" StatementType="SELECT" RetrievedFromCache="true" StatementSubTreeCost="0.00657038" StatementEstRows="1" StatementOptmLevel="FULL" QueryHash="0x7E06C77E90EB9FBB" QueryPlanHash="0x64478FC6152D2A83" StatementOptmEarlyAbortReason="GoodEnoughPlanFound" CardinalityEstimationModelVersion="120" ParameterizedText="(@1 tinyint)SELECT * FROM [EmployeeOp] WHERE [VacationHours]>@1">
<StatementSetOptions QUOTED_IDENTIFIER="true" ARITHABORT="true" CONCAT_NULL_YIELDS_NULL="true" ANSI_NULLS="true" ANSI_PADDING="true" ANSI_WARNINGS="true" NUMERIC_ROUNDABORT="false" />
<QueryPlan CachedPlanSize="40" CompileTime="6" CompileCPU="6" CompileMemory="232">
<MemoryGrantInfo SerialRequiredMemory="0" SerialDesiredMemory="0" />
<OptimizerHardwareDependentProperties EstimatedAvailableMemoryGrant="206857" EstimatedPagesCached="51714" EstimatedAvailableDegreeOfParallelism="2" />
<RelOp NodeId="0" PhysicalOp="Nested Loops" LogicalOp="Inner Join" EstimateRows="1" EstimateIO="0" EstimateCPU="4.18e-006" AvgRowSize="830" EstimatedTotalSubtreeCost="0.00657038" Parallel="0" EstimateRebinds="0" EstimateRewinds="0" EstimatedExecutionMode="Row">
<OutputList>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="BusinessEntityID" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="NationalIDNumber" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="LoginID" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="OrganizationNode" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="OrganizationLevel" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="JobTitle" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="BirthDate" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="MaritalStatus" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="Gender" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="HireDate" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="SalariedFlag" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="VacationHours" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="SickLeaveHours" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="CurrentFlag" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="rowguid" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="ModifiedDate" />
</OutputList>
<NestedLoops Optimized="0">
<OuterReferences>
<ColumnReference Column="Bmk1000" />
</OuterReferences>
<RelOp NodeId="1" PhysicalOp="Index Seek" LogicalOp="Index Seek" EstimateRows="1" EstimateIO="0.003125" EstimateCPU="0.0001581" AvgRowSize="19" EstimatedTotalSubtreeCost="0.0032831" TableCardinality="290" Parallel="0" EstimateRebinds="0" EstimateRewinds="0" EstimatedExecutionMode="Row">
<OutputList>
<ColumnReference Column="Bmk1000" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="VacationHours" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="SickLeaveHours" />
</OutputList>
<IndexScan Ordered="1" ScanDirection="FORWARD" ForcedIndex="0" ForceSeek="0" ForceScan="0" NoExpandHint="0" Storage="RowStore">
<DefinedValues>
<DefinedValue>
<ColumnReference Column="Bmk1000" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="VacationHours" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="SickLeaveHours" />
</DefinedValue>
</DefinedValues>
<Object Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Index="[nc_Employee_vacationsickleave]" IndexKind="NonClustered" Storage="RowStore" />
<SeekPredicates>
<SeekPredicateNew>
<SeekKeys>
<StartRange ScanType="GT">
<RangeColumns>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="VacationHours" />
</RangeColumns>
<RangeExpressions>
<ScalarOperator ScalarString="(99)">
<Const ConstValue="(99)" />
</ScalarOperator>
</RangeExpressions>
</StartRange>
</SeekKeys>
</SeekPredicateNew>
</SeekPredicates>
</IndexScan>
</RelOp>
<RelOp NodeId="3" PhysicalOp="RID Lookup" LogicalOp="RID Lookup" EstimateRows="1" EstimateIO="0.003125" EstimateCPU="0.0001581" AvgRowSize="826" EstimatedTotalSubtreeCost="0.0032831" TableCardinality="290" Parallel="0" EstimateRebinds="0" EstimateRewinds="0" EstimatedExecutionMode="Row">
<OutputList>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="BusinessEntityID" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="NationalIDNumber" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="LoginID" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="OrganizationNode" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="OrganizationLevel" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="JobTitle" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="BirthDate" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="MaritalStatus" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="Gender" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="HireDate" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="SalariedFlag" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="CurrentFlag" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="rowguid" />
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="ModifiedDate" />
</OutputList>
<IndexScan Lookup="1" Ordered="1" ScanDirection="FORWARD" ForcedIndex="0" ForceSeek="0" ForceScan="0" NoExpandHint="0" Storage="RowStore">
<DefinedValues>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="BusinessEntityID" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="NationalIDNumber" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="LoginID" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="OrganizationNode" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="OrganizationLevel" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="JobTitle" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="BirthDate" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="MaritalStatus" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="Gender" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="HireDate" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="SalariedFlag" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="CurrentFlag" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="rowguid" />
</DefinedValue>
<DefinedValue>
<ColumnReference Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" Column="ModifiedDate" />
</DefinedValue>
</DefinedValues>
<Object Database="[HRDB]" Schema="[dbo]" Table="[EmployeeOp]" TableReferenceId="-1" IndexKind="Heap" Storage="RowStore" />
<SeekPredicates>
<SeekPredicateNew>
<SeekKeys>
<Prefix ScanType="EQ">
<RangeColumns>
<ColumnReference Column="Bmk1000" />
</RangeColumns>
<RangeExpressions>
<ScalarOperator ScalarString="[Bmk1000]">
<Identifier>
<ColumnReference Column="Bmk1000" />
</Identifier>
</ScalarOperator>
</RangeExpressions>
</Prefix>
</SeekKeys>
</SeekPredicateNew>
</SeekPredicates>
</IndexScan>
</RelOp>
</NestedLoops>
</RelOp>
<ParameterList>
<ColumnReference Column="@1" ParameterCompiledValue="(99)" />
</ParameterList>
</QueryPlan>
</StmtSimple>
</Statements>
</Batch>
</BatchSequence>
</ShowPlanXML>
\') --表扫描
select * from sys.dm_exec_cached_plans
select * from sys.dm_exec_sql_text
select * from sys.dm_exec_query_plan
select * from EmployeeOp where VacationHours>99 -- 索引
select c.plan_handle,p.text from sys.dm_exec_cached_plans as c cross apply sys.dm_exec_sql_text(c.plan_handle) as p
select * from sys.dm_exec_query_plan(0x06000A00CD253E14207CA0290200000001000000000000000000000000000000000000000000000000000000)
以上是关于SqlServer性能优化 即席查询（十三）的主要内容，如果未能解决你的问题，请参考以下文章