最近在忙着优化集团公司的一个报表。优化完成后，报表查询速度由从半小时以上（甚至查不出）到秒查的质变。从修改 SQL 查询语句逻辑到决定创建存储过程实现，花了我 3 天多的时间，在此总结一下，希望对朋友们有帮助。

数据背景

首先项目是西门子中国在我司实施部署的 MES 项目，由于项目是在产线上运作(3 years+)，数据累积很大。

在项目的数据库中，大概上亿条数据的表有 5 个以上，千万级数据的表 10 个以上，百万级数据的表，很多…

（历史问题，当初实施无人监管，无人监控数据库这块的性能问题。PS：我刚入职不久…）

不多说，直接贴西门子中国的开发人员在我司开发的 SSRS 报表中的 SQL 语句：

select distinct b.MaterialID as matl_def_id, c.Descript, case when right(b.MESOrderID, 12) < '001000000000' then right(b.MESOrderID, 9)
else right(b.MESOrderID, 12) end  as pom_order_id, a.LotName, a.SourceLotName as ComLot,
e.DefID as ComMaterials, e.Descript as ComMatDes, d.VendorID, d.DateCode,d.SNNote, b.OnPlantID,a.SNCUST
from
(select m.lotname, m.sourcelotname, m.opetypeid, m.OperationDate,n.SNCUST from View1 mleft join co_sn_link_customer as n on n.SNMes=m.LotNamewhere( m.LotName in (select val from fn_String_To_Table(@sn,',',1)) or (@sn) = '') and( m.sourcelotname in (select val from fn_String_To_Table(@BatchID,',',1)) or (@BatchID) = '')and (n.SNCust like '%'+ @SN_ext + '%' or (@SN_ext)='')
) a
left join
(select * from Table1 where SNType = 'IntSN'and SNRuleName = 'ProductSNRule'and OnPlantID=@OnPlant
) b on b.SN = a.LotName
inner join MMdefinitions as c on c.DefID = b.MaterialID
left join  Table1 as d on d.SN = a.SourceLotName
inner join MMDefinitions as e on e.DefID = d.MaterialID
where not exists (select distinct LotName, SourceLotName from ELCV_ASSEMBLE_OPS
where LotName = a.SourceLotName and SourceLotName = a.LotName
)
and (d.DateCode in (select val from fn_String_To_Table(@DCode,',',1)) or (@DCode) = '')
and (d.SNNote  like '%'+@SNNote+'%' or (@SNNote) = '')
and ((case when right(b.MESOrderID, 12) < '001000000000' then right(b.MESOrderID, 9)
else right(b.MESOrderID, 12) end) in (select val from fn_String_To_Table(@order_id,',',1)) or (@order_id) = '')
and (e.DefID in (select val from fn_String_To_Table(@comdef,',',1)) or (@comdef) = '')
--View1是一个嵌套两层的视图(出于保密性，实际名称可能不同)，里面有一张上亿数据的表和几张千万级数据的表做左连接查询
--Table1是一个数据记录超过1500万的表

这个查询语句，实际上通过我的检测和调查，在 B/S 系统前端已无法查出结果，半小时，一小时...

因为我直接在 SQL 查询分析器查，半小时都没有结果。（原因是里面对一张上亿级数据表和 3 张千万级数据表做全表扫描查询）

不由感慨，西门子中国的素质（或者说责任感）就这样？下面说说我的分析和走的弯路（思维误区），希望对你也有警醒。

“

推荐一个艿艿写的 6000+ Star 的 SpringBoot + SpringCloud + Dubbo 教程的仓库：https://github.com/YunaiV/SpringBoot-Labs

探索和误区

首先相关表的索引，没有建全的，把索引给建上。

索引这步完成后，发现情况还是一样，查询速度几乎没有改善。后来想起相关千万级数据以上的表，都还没有建立表分区。于是考虑建立表分区以及数据复制的方案。

这里有必要说明下： 我司报表用的是一个专门的数据库服务器，数据从产线订阅而来。就是常说的“读写分离”。

如果直接在原表上建立表分区，你会发现执行表分区的事物会直接死锁。原因是：表分区操作本身会锁表，产线还在推数据过来，这样很容易“阻塞”，“死锁”。

我想好的方案是：建立一个新表（空表），在新表上建好表分区，然后复制数据过来。

正打算这么干。等等！我好像进入了一个严重的误区！

分析： 原 SQL 语句和业务需求，是对产线的数据做产品以及序列号的追溯，关键是查询条件里没有有规律的”条件”（如日期、编号），贸然做了表分区，在这里几乎没有意义！反而会降低查询性能！

好险！还是一步一步来，先做 SQL 语句分析。

对原 SQL 语句的分析

对原 SQL 语句的分析如下：

查询语句的 where 条件，有大量 @var in … or (@var =”) 的片段。
where 条件有 like ‘%’+@var+’%’。
where 条件有 case … end 函数。
多次连接同一表查询，另外使用本身已嵌套的视图表，是不是必须，是否可替代？
SQL 语句有号，视图中也有号出现。

优化设计

首先是用存储过程改写，好处是设计灵活。

核心思想是： 用一个或多个查询条件（查询条件要求至少输入一个）得到临时表，每个查询条件如果查到集合，就更新这张临时表，最后汇总的时候，只需判断这个临时表是否有值。

以此类推，可以建立多个临时表，将查询条件汇总。

这样做目前来看至少两点好处：

省去了对变量进行 =@var or (@var=”)的判断。
抛弃 SQL 拼接，提高代码可读性。

再有就是在书写存储过程，这个过程中要注意：

尽量想办法使用临时表扫描替代全表扫描。
抛弃 in 和 not in 语句，使用 exists 和 not exists 替代。
和客户确认，模糊查询是否有必要，如没有必要，去掉 like 语句。
注意建立适当的，符合场景的索引。
踩死 “*” 号。
避免在 where 条件中对字段进行函数操作。
对实时性要求不高的报表，允许脏读（with(nolock)）。

“

推荐一个艿艿写的 3000+ Star 的 SpringCloud Alibaba 电商开源项目的仓库：https://github.com/YunaiV/onemall

存储过程

如果想参考优化设计片段的详细内容，请参阅 SQL 代码：

/*** 某某跟踪报表**/
--exec spName1 '','','','','','','公司代号'
CREATE Procedure spName1@MESOrderID nvarchar(320), --工单号,最多30个@LotName nvarchar(700),    --产品序列号,最多50个@DateCode nvarchar(500),   --供应商批次号,最多30个@BatchID nvarchar(700),    --组装件序列号/物料批号,最多50个@comdef nvarchar(700),     --组装件物料编码,最多30个@SNCust nvarchar(1600),    --外部序列号,最多50个@OnPlant nvarchar(20)      --平台
AS
BEGINSET NOCOUNT ON;/*** 1)定义全局的临时表，先根据六个查询条件的任意一个，得出临时表结果**/CREATE TABLE #FinalLotName(LotName NVARCHAR(50),       --序列号SourceLotName NVARCHAR(50), --来源序列号SNCust NVARCHAR(128)        --外部序列号)--1.1IF @LotName<>''BEGINSELECT Val INTO #WorkLot FROM fn_String_To_Table(@LotName,',',1)SELECT LotPK,LotName INTO #WorkLotPK FROM MMLots WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #WorkLot b WHERE b.Val=MMLots.LotID)--求SourceLotPK只能在这里求SELECT a.LotPK,a.SourceLotPK into #WorkSourcePK FROM MMLotOperations a WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #WorkLotPK b WHERE b.LotPK=a.LotPK) AND a.SourceLotPK IS NOT NULLSELECT a.LotPK,a.SourceLotPK,b.LotName INTO #WorkSourcePK2 FROM #WorkSourcePK a JOIN #WorkLotPK b ON a.LotPK=b.LotPKINSERT INTO #FinalLotName SELECT a.LotName,b.LotName AS SourceLotName,NULL FROM #WorkSourcePK2 a JOIN (SELECT LotPK,LotName FROM MMLots WITH(NOLOCK) ) b on a.SourceLotPK=b.LotPK --b的里面加不加WHERE RowDeleted=0待确定SELECT a.LotName,a.SourceLotName,b.SNCust INTO #FinalLotNameX1 FROM #FinalLotName a LEFT JOIN CO_SN_LINK_CUSTOMER b WITH(NOLOCK) ON a.LotName=b.SNMesDELETE FROM #FinalLotNameINSERT INTO #FinalLotName SELECT LotName,SourceLotName,SNCust FROM #FinalLotNameX1END--1.2IF @BatchID<>''BEGINSELECT Val INTO #WorkSourceLot FROM fn_String_To_Table(@BatchID,',',1)IF EXISTS(SELECT 1 FROM #FinalLotName)--如果@LotName也不为空BEGINSELECT a.LotName,a.SourceLotName,a.SNCust INTO #FinalLotNameX2 FROM #FinalLotName a WHERE EXISTS(SELECT 1 FROM #WorkSourceLot b WHERE a.SourceLotName=b.Val)DELETE FROM #FinalLotNameINSERT INTO #FinalLotName SELECT LotName,SourceLotName,SNCust FROM #FinalLotNameX2ENDELSE --@LotName条件为空BEGINSELECT LotPK AS SourceLotPK,LotName AS SourceLotName INTO #2 FROM MMLots WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #WorkSourceLot b WHERE b.Val=MMLots.LotID)SELECT a.LotPK,a.SourceLotPK into #21 FROM MMLotOperations a WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #2 b WHERE b.SourceLotPK=a.SourceLotPK)SELECT a.LotPK,a.SourceLotPK,b.SourceLotName INTO #22 FROM #21 a JOIN #2 b ON a.SourceLotPK=b.SourceLotPKINSERT INTO #FinalLotName SELECT b.LotName,a.SourceLotName,NULL FROM #22 a JOIN (SELECT LotPK,LotName FROM MMLots WITH(NOLOCK) ) b on a.LotPK=b.LotPK --b的里面加不加WHERE RowDeleted=0待确定SELECT a.LotName,a.SourceLotName,b.SNCust INTO #FinalLotNameX21 FROM #FinalLotName a LEFT JOIN CO_SN_LINK_CUSTOMER b WITH(NOLOCK) ON a.LotName=b.SNMesDELETE FROM #FinalLotNameINSERT INTO #FinalLotName SELECT LotName,SourceLotName,SNCust FROM #FinalLotNameX21ENDEND--1.3IF @SNCust<>''BEGINSELECT Val INTO #WorkCustomSN FROM fn_String_To_Table(@SNCust,',',1)IF EXISTS(SELECT 1 FROM #FinalLotName)--前面两个条件至少有一个有值BEGINSELECT a.LotName,a.SourceLotName,a.SNCust INTO #FinalLotNameX3 FROM #FinalLotName a WHERE EXISTS(SELECT 1 FROM #WorkCustomSN b WHERE a.SNCust=b.Val)DELETE FROM #FinalLotNameINSERT INTO #FinalLotName SELECT LotName,SourceLotName,SNCust FROM #FinalLotNameX3ENDELSEBEGINSELECT a.SNMes INTO #WorkLotX FROM CO_SN_LINK_CUSTOMER a WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #WorkCustomSN b WHERE a.SNCust=b.Val)-------------------以下逻辑和变量1(@LotName)类似[先根据外部序列号求解序列号,再照搬第一个判断变量的方式]SELECT LotPK,LotName INTO #WorkLotPKX FROM MMLots WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #WorkLotX b WHERE b.SNMes=MMLots.LotID)--求SourceLotPK只能在这里求SELECT a.LotPK,a.SourceLotPK into #WorkSourcePKX FROM MMLotOperations a WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #WorkLotPKX b WHERE b.LotPK=a.LotPK) AND a.SourceLotPK IS NOT NULLSELECT a.LotPK,a.SourceLotPK,b.LotName INTO #WorkSourcePK2X FROM #WorkSourcePKX a JOIN #WorkLotPKX b ON a.LotPK=b.LotPKINSERT INTO #FinalLotName SELECT a.LotName,b.LotName AS SourceLotName,NULL FROM #WorkSourcePK2X a JOIN (SELECT LotPK,LotName FROM MMLots WITH(NOLOCK) ) b on a.SourceLotPK=b.LotPK --b的里面加不加WHERE RowDeleted=0待确定SELECT a.LotName,a.SourceLotName,b.SNCust INTO #FinalLotNameX31 FROM #FinalLotName a LEFT JOIN CO_SN_LINK_CUSTOMER b WITH(NOLOCK) ON a.LotName=b.SNMesDELETE FROM #FinalLotNameINSERT INTO #FinalLotName SELECT LotName,SourceLotName,SNCust FROM #FinalLotNameX31-----------------------ENDEND/*** 2)定义全局的临时表，用于替换第一个全局临时表。**/CREATE TABLE #FinalCO_SN(SN NVARCHAR(50),SourceSN NVARCHAR(50),SNCust NVARCHAR(128),matl_def_id NVARCHAR(50),--sn的物料IDComMaterials NVARCHAR(50),  --SourceSN的物料IDMESOrderID NVARCHAR(20),OnPlantID NVARCHAR(20),VendorID NVARCHAR(20),DateCode NVARCHAR(20) ,SNNote NVARCHAR(512))--2.1IF @MESOrderID<>''BEGIN-------------------------------将MESOrderID做特殊处理-----------------------------------SELECT Val INTO #WorkMESOrderID FROM fn_String_To_Table(@MESOrderID,',',1)IF @OnPlant='Comba'BEGINUPDATE #WorkMESOrderID SET Val='C000'+Val WHERE LEN(Val)=9ENDELSEBEGINUPDATE #WorkMESOrderID SET Val='W000'+Val WHERE LEN(Val)=9ENDSELECT SN,MaterialID,MESOrderID,OnPlantID INTO #WorkCO_SN1 FROM CO_SN_GENERATION a WITH(NOLOCK)WHERE SNType='IntSN' AND SNRuleName = 'ProductSNRule' AND OnPlantID=@OnPlantAND EXISTS(SELECT 1 FROM #WorkMESOrderID b WHERE a.MESOrderID=b.Val)--------------------------------------------------------------------------------------------条件判断(逻辑分析)开始IF EXISTS(SELECT 1 FROM #FinalLotName)--如果前面判断的查询条件有值BEGIN--查出SourceLotName对应的查询字段SELECT a.SN AS SourceLotName,a.VendorID,a.DateCode,a.SNNote,a.MaterialID AS ComMaterials INTO #SourceLotNameTable FROM CO_SN_GENERATION a WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #FinalLotName b WHERE a.SN=b.SourceLotName)INSERT INTO #FinalCO_SNSELECT a.LotName,a.SourceLotName,d.SNCust,b.MaterialID,c.ComMaterials,b.MESOrderID,b.OnPlantID,c.VendorID,c.DateCode,c.SNNote FROM #FinalLotName aLEFT JOIN #WorkCO_SN1 b ON a.LotName=b.SNLEFT JOIN #SourceLotNameTable c ON a.SourceLotName=c.SourceLotNameLEFT JOIN CO_SN_LINK_CUSTOMER d WITH(NOLOCK) ON a.LotName=d.SNMesENDELSEBEGIN--已知SN集合求解对应的SourceSN和SNCust集合------------------------------------------SELECT LotPK,LotName INTO #WorkLotPK410 FROM MMLots WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #WorkCO_SN1 b WHERE b.SN=MMLots.LotID)SELECT a.LotPK,a.SourceLotPK into #WorkSourcePK420 FROM MMLotOperations a WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #WorkLotPK410 b WHERE b.LotPK=a.LotPK) AND a.SourceLotPK IS NOT NULLSELECT a.LotPK,a.SourceLotPK,b.LotName INTO #WorkSourcePK430 FROM #WorkSourcePK420 a JOIN #WorkLotPK410 b ON a.LotPK=b.LotPKINSERT INTO #FinalLotName SELECT a.LotName,b.LotName AS SourceLotName,NULL FROM #WorkSourcePK430 a JOIN (SELECT LotPK,LotName FROM MMLots WITH(NOLOCK) ) b on a.SourceLotPK=b.LotPK --b的里面加不加WHERE RowDeleted=0待确定SELECT a.LotName,a.SourceLotName,b.SNCust INTO #FinalLotNameX440 FROM #FinalLotName a LEFT JOIN CO_SN_LINK_CUSTOMER b WITH(NOLOCK) ON a.LotName=b.SNMesDELETE FROM #FinalLotNameINSERT INTO #FinalLotName SELECT LotName,SourceLotName,SNCust FROM #FinalLotNameX440-------------------------------------------------------------------------------------SELECT a.SN AS SourceLotName,a.VendorID,a.DateCode,a.SNNote,a.MaterialID AS ComMaterials INTO #SourceLotNameTable2 FROM CO_SN_GENERATION a WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #FinalLotName b WHERE a.SN=b.SourceLotName)INSERT INTO #FinalCO_SNSELECT a.LotName,a.SourceLotName,a.SNCust,b.MaterialID,c.ComMaterials,b.MESOrderID,b.OnPlantID,c.VendorID,c.DateCode,c.SNNote FROM #FinalLotName aLEFT JOIN #WorkCO_SN1 b ON a.LotName=b.SNLEFT JOIN #SourceLotNameTable2 c ON a.SourceLotName=c.SourceLotNameENDEND--2.2IF @DateCode<>''BEGINSELECT Val INTO #WorkDateCode FROM fn_String_To_Table(@DateCode,',',1)--此@DataCode条件求解出来的是SourceSNSELECT SN AS SourceSN,MaterialID AS ComMaterials,VendorID,DateCode,SNNote INTO #WorkSourceSNT1 FROM CO_SN_GENERATION a WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #WorkDateCode b WHERE a.DateCode=b.Val)------------------------------------------------------------------------------------------------------条件判断(逻辑分析)开始IF EXISTS(SELECT 1 FROM #FinalCO_SN)--如果前面判断的查询条件有值BEGINSELECT a.LotName,a.SourceLotName,a.SNCust,a.MaterialID,a.ComMaterials,a.MESOrderID,a.OnPlantID,a.VendorID,a.DateCode,a.SNNote INTO #TMP51 FROM #FinalCO_SN a WHERE EXISTS (SELECT 1 FROM #WorkDateCode b WHERE a.DateCode=b.Val)DELETE FROM #FinalCO_SNINSERT INTO #FinalCO_SN SELECT LotName,SourceLotName,SNCust,MaterialID,ComMaterials,MESOrderID,OnPlantID,VendorID,DateCode,SNNote FROM #TMP51ENDELSEBEGINIF EXISTS(SELECT 1 FROM #FinalLotName)BEGIN--查出SourceLotName对应的查询字段SELECT a.SourceSN,a.VendorID,a.DateCode,a.SNNote,a.ComMaterials INTO #SourceLTX5 FROM #WorkSourceSNT1 a WHERE EXISTS(SELECT 1 FROM #FinalLotName b WHERE a.SourceSN=b.SourceLotName)--查出SN对应的查询字段SELECT SN,MaterialID,MESOrderID,OnPlantID INTO #WorkSNT510 FROM CO_SN_GENERATION a WITH(NOLOCK)WHERE SNType='IntSN' AND SNRuleName = 'ProductSNRule' AND OnPlantID=@OnPlantAND EXISTS(SELECT 1 FROM #FinalLotName b WHERE a.SN=b.LotName)INSERT INTO #FinalCO_SNSELECT a.LotName,a.SourceLotName,d.SNCust,b.MaterialID,c.ComMaterials,b.MESOrderID,b.OnPlantID,c.VendorID,c.DateCode,c.SNNote FROM #FinalLotName aLEFT JOIN #WorkSNT510 b ON a.LotName=b.SNLEFT JOIN #WorkSourceSNT1 c ON a.SourceLotName=c.SourceSNLEFT JOIN CO_SN_LINK_CUSTOMER d WITH(NOLOCK) ON a.LotName=d.SNMesENDELSEBEGIN--已知SourceSN集合求解对应的SN和SNCust集合------------------------------------------SELECT LotPK AS SourceLotPK,LotName AS SrouceLotName INTO #WorkLotX510 FROM MMLots WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #WorkSourceSNT1 b WHERE b.SourceSN=MMLots.LotID)SELECT a.LotPK,a.SourceLotPK into #WorkLotX520 FROM MMLotOperations a WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #WorkLotX510 b WHERE b.SourceLotPK=a.SourceLotPK)SELECT a.LotPK,a.SourceLotPK,b.SrouceLotName INTO #WorkLotX530 FROM #WorkLotX520 a JOIN #WorkLotX510 b ON a.SourceLotPK=b.SourceLotPKINSERT INTO #FinalLotName SELECT b.LotName,a.SrouceLotName,NULL FROM #WorkLotX530 a JOIN (SELECT LotPK,LotName FROM MMLots WITH(NOLOCK) ) b on a.LotPK=b.LotPK --b的里面加不加WHERE RowDeleted=0待确定SELECT a.LotName,a.SourceLotName,b.SNCust INTO #WorkLotX540 FROM #FinalLotName a LEFT JOIN CO_SN_LINK_CUSTOMER b WITH(NOLOCK) ON a.LotName=b.SNMesDELETE FROM #FinalLotNameINSERT INTO #FinalLotName SELECT LotName,SourceLotName,SNCust FROM #WorkLotX540-------------------------------------------------------------------------------------SELECT SN,MaterialID,MESOrderID,OnPlantID INTO #WorkLotX550 FROM CO_SN_GENERATION a WITH(NOLOCK)WHERE SNType='IntSN' AND SNRuleName = 'ProductSNRule' AND OnPlantID=@OnPlantAND EXISTS(SELECT 1 FROM #FinalLotName b WHERE a.SN=b.LotName)INSERT INTO #FinalCO_SNSELECT a.LotName,a.SourceLotName,a.SNCust,b.MaterialID,c.ComMaterials,b.MESOrderID,b.OnPlantID,c.VendorID,c.DateCode,c.SNNote FROM #FinalLotName aLEFT JOIN #WorkLotX550 b ON a.LotName=b.SNLEFT JOIN #WorkSourceSNT1 c ON a.SourceLotName=c.SourceSNENDENDEND--2.3IF @comdef<>''BEGINSELECT Val INTO #WorkComdef FROM fn_String_To_Table(@comdef,',',1)--此@comdef条件求解出来的是SourceSNSELECT SN AS SourceSN,MaterialID AS ComMaterials,VendorID,DateCode,SNNote INTO #WorkSourceSNT16 FROM CO_SN_GENERATION a WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #WorkComdef b WHERE a.MaterialID=b.Val)------------------------------------------------------------------------------------------------------条件判断(逻辑分析)开始IF EXISTS(SELECT 1 FROM #FinalCO_SN)--如果前面判断的查询条件有值BEGINSELECT a.LotName,a.SourceLotName,a.SNCust,a.MaterialID,a.ComMaterials,a.MESOrderID,a.OnPlantID,a.VendorID,a.DateCode,a.SNNote INTO #TMP516 FROM #FinalCO_SN a WHERE EXISTS (SELECT 1 FROM #WorkComdef b WHERE a.matl_def_id=b.Val)DELETE FROM #FinalCO_SNINSERT INTO #FinalCO_SN SELECT LotName,SourceLotName,SNCust,MaterialID,ComMaterials,MESOrderID,OnPlantID,VendorID,DateCode,SNNote FROM #TMP516ENDELSEBEGINIF EXISTS(SELECT 1 FROM #FinalLotName)BEGIN--查出SourceLotName对应的查询字段SELECT a.SourceSN,a.VendorID,a.DateCode,a.SNNote,a.ComMaterials INTO #SourceLTX56 FROM #WorkSourceSNT16 a WHERE EXISTS(SELECT 1 FROM #FinalLotName b WHERE a.SourceSN=b.SourceLotName)--查出SN对应的查询字段SELECT SN,MaterialID,MESOrderID,OnPlantID INTO #WorkSNT5106 FROM CO_SN_GENERATION a WITH(NOLOCK)WHERE SNType='IntSN' AND SNRuleName = 'ProductSNRule' AND OnPlantID=@OnPlantAND EXISTS(SELECT 1 FROM #FinalLotName b WHERE a.SN=b.LotName)INSERT INTO #FinalCO_SNSELECT a.LotName,a.SourceLotName,d.SNCust,b.MaterialID,c.ComMaterials,b.MESOrderID,b.OnPlantID,c.VendorID,c.DateCode,c.SNNote FROM #FinalLotName aLEFT JOIN #WorkSNT5106 b ON a.LotName=b.SNLEFT JOIN #WorkSourceSNT16 c ON a.SourceLotName=c.SourceSNLEFT JOIN CO_SN_LINK_CUSTOMER d WITH(NOLOCK) ON a.LotName=d.SNMesENDELSEBEGIN--已知SourceSN集合求解对应的SN和SNCust集合------------------------------------------SELECT LotPK AS SourceLotPK,LotName AS SrouceLotName INTO #WorkLotX5106 FROM MMLots WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #WorkSourceSNT16 b WHERE b.SourceSN=MMLots.LotID)SELECT a.LotPK,a.SourceLotPK into #WorkLotX5206 FROM MMLotOperations a WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #WorkLotX5106 b WHERE b.SourceLotPK=a.SourceLotPK)SELECT a.LotPK,a.SourceLotPK,b.SrouceLotName INTO #WorkLotX5306 FROM #WorkLotX5206 a JOIN #WorkLotX5106 b ON a.SourceLotPK=b.SourceLotPKINSERT INTO #FinalLotName SELECT b.LotName,a.SrouceLotName,NULL FROM #WorkLotX5306 a JOIN (SELECT LotPK,LotName FROM MMLots WITH(NOLOCK) ) b on a.LotPK=b.LotPK --b的里面加不加WHERE RowDeleted=0待确定SELECT a.LotName,a.SourceLotName,b.SNCust INTO #WorkLotX5406 FROM #FinalLotName a LEFT JOIN CO_SN_LINK_CUSTOMER b WITH(NOLOCK) ON a.LotName=b.SNMesDELETE FROM #FinalLotNameINSERT INTO #FinalLotName SELECT LotName,SourceLotName,SNCust FROM #WorkLotX5406-------------------------------------------------------------------------------------SELECT SN,MaterialID,MESOrderID,OnPlantID INTO #WorkLotX5506 FROM CO_SN_GENERATION a WITH(NOLOCK)WHERE SNType='IntSN' AND SNRuleName = 'ProductSNRule' AND OnPlantID=@OnPlantAND EXISTS(SELECT 1 FROM #FinalLotName b WHERE a.SN=b.LotName)INSERT INTO #FinalCO_SNSELECT a.LotName,a.SourceLotName,a.SNCust,b.MaterialID,c.ComMaterials,b.MESOrderID,b.OnPlantID,c.VendorID,c.DateCode,c.SNNote FROM #FinalLotName aLEFT JOIN #WorkLotX5506 b ON a.LotName=b.SNLEFT JOIN #WorkSourceSNT16 c ON a.SourceLotName=c.SourceSNENDENDEND/*** 3)条件判断结束**/IF EXISTS(SELECT 1 FROM #FinalLotName)BEGINIF EXISTS(SELECT 1 FROM #FinalCO_SN)BEGIN--3.1SELECT a.matl_def_id,b.Descript,a.MESOrderID AS pom_order_id,a.SN AS LotName,a.SourceSN AS ComLot,a.ComMaterials,c.Descript AS ComMatDes,a.VendorID,a.DateCode,a.SNNote,OnPlantID,SNCust FROM #FinalCO_SN aJOIN MMDefinitions b WITH(NOLOCK) ON a.matl_def_id=b.DefIDJOIN MMDefinitions c WITH(NOLOCK) ON a.ComMaterials=c.DefIDWHERE NOT EXISTS(select distinct SN, SourceSN from #FinalCO_SN xwhere x.SN = a.SourceSN and x.SourceSN = a.SN)ENDELSEBEGIN--3.2--3.2.1求解SN的必查字段SELECT SN,MaterialID,MESOrderID,OnPlantID INTO #FinalSNX1 FROM CO_SN_GENERATION a WITH(NOLOCK)WHERE SNType='IntSN' AND SNRuleName = 'ProductSNRule' AND OnPlantID=@OnPlantAND EXISTS(SELECT 1 FROM #FinalLotName b WHERE a.SN=b.LotName)--3.2.2求解SourceSN的必查字段SELECT a.SN AS SourceLotName,a.VendorID,a.DateCode,a.SNNote,a.MaterialID AS ComMaterials INTO #FinalSNX2 FROM CO_SN_GENERATION a WITH(NOLOCK) WHERE EXISTS(SELECT 1 FROM #FinalLotName b WHERE a.SN=b.SourceLotName)SELECT b.MaterialID AS matl_def_id,x.Descript,b.MESOrderID AS pom_order_id,b.SN AS LotName,c.SourceLotName AS ComLot,c.ComMaterials,y.Descript AS ComMatDes,c.VendorID,c.DateCode,c.SNNote,b.OnPlantID,a.SNCustFROM #FinalLotName aLEFT JOIN #FinalSNX1 b ON a.LotName=b.SNLEFT JOIN #FinalSNX2 c ON a.SourceLotName=c.SourceLotNameJOIN MMDefinitions x WITH(NOLOCK) ON b.MaterialID=x.DefIDJOIN MMDefinitions y WITH(NOLOCK) ON c.ComMaterials=y.DefIDWHERE NOT EXISTS(SELECT DISTINCT * FROM #FinalLotName zWHERE z.LotName=a.SourceLotName and z.SourceLotName=a.LotName)ENDENDELSEBEGINIF EXISTS(SELECT 1 FROM #FinalCO_SN)BEGIN--3.3SELECT a.matl_def_id,b.Descript,a.MESOrderID AS pom_order_id,a.SN AS LotName,a.SourceSN AS ComLot,a.ComMaterials,c.Descript AS ComMatDes,a.VendorID,a.DateCode,a.SNNote,OnPlantID,SNCust FROM #FinalCO_SN aJOIN MMDefinitions b WITH(NOLOCK) ON a.matl_def_id=b.DefIDJOIN MMDefinitions c WITH(NOLOCK) ON a.ComMaterials=c.DefIDWHERE NOT EXISTS(select distinct SN, SourceSN from #FinalCO_SN xwhere x.SN = a.SourceSN and x.SourceSN = a.SN)ENDELSEBEGIN--3.4PRINT 'There is no queryable condition,please enter at less a query conditon.'ENDENDEND
GO

虽然牺牲了代码的可读性，但创造了性能价值。本人水平有限，还请各位不吝赐教！

最后，将 SSRS 报表替换成此存储过程后，SQL 查询分析器是秒查的。B/S 前端用时 1~2 秒！

总结

平常的你是否偶尔会因急于完成任务而书写一堆性能极低的 SQL 语句呢？写出可靠性能的 SQL 语句不难，难的是习惯。

本文的优化思想很简单，关键点是避免全表扫描&注重 SQL 语句写法&索引。

另外，如果你查询的表有可能会在查询时段更新，而实际业务需求允许脏读，可加 with（nolock）预防查询被更新事物阻塞。

耗时 3 天，上亿数据如何做到秒级查询？相关推荐

耗时3天，上亿数据如何做到秒级查询？
点击上方"方志朋",选择"设为星标" 回复"666"获取新整理的面试文章来源:sohu.gg/jIp59N 最近在忙着优化集团公司的一个报 ...
es对几十亿数据能达到秒级响应吗_万亿数据下的多维实时分析系统，如何做到亚秒级响应...
导语当业务发展到一定规模,实时数据仓库是一个必要的基础服务.从数据驱动方面考虑,多维实时数据分析系统的重要性也不言而喻.但是当数据量巨大的情况下,拿腾讯看点来说,一天上报的数据量达到万亿级的规模,要 ...
上亿数据怎么玩深度分页？兼容MySQL + ES + MongoDB
面试题 & 真实经历面试题:在数据量很大的情况下,怎么实现深度分页? 大家在面试时,或者准备面试中可能会遇到上述的问题,大多的回答基本上是分库分表建索引,这是一种很标准的正确回答,但现实总是 ...
查询a表有但是b表没有的数据_牛逼！一个上亿数据的报表竟然能做到秒查~
数据背景首先项目是西门子中国在我司实施部署的MES项目,由于项目是在产线上运作(3 years+),数据累积很大.在项目的数据库中,大概上亿条数据的表有5个以上,千万级数据的表10个以上,百万级数据 ...
实战上亿数据，如何实现秒查！
点击上方好好学java ,选择星标公众号重磅资讯.干货,第一时间送达今日推荐:硬刚一周,3W字总结,一年的经验告诉你如何准备校招! 个人原创100W+访问量博客:点击前往,查看更多作者:J ...
前端如何展示一个有上亿数据的树结构
需求描述数据上亿,有数千个从属于根节点的一级子节点: 每个一级子节点有数个到数万个二级子节点: 二级子节点可能有三级子节点: 子节点的最多层深不定(即,也许可能有 10 层): 需求分析当面临一个 ...
实战：上亿数据如何秒查
作者:blog.csdn.net/chenleixing/article/details/44994571 最近在忙着优化集团公司的一个报表.优化完成后,报表查询速度有从半小时以上(甚至查不出)到秒查 ...
高效率批量插入上亿数据
转至 : https://blog.csdn.net/xiyang_1990/article/details/78771962 这里提供一种方法,使用 APPEND 提示,使得十分钟内插入上亿数据成为 ...
双引擎驱动Quick BI十亿数据0.3秒分析，首屏展示时间缩短30%
简介:在规划中,Quick BI制定了产品竞争力建设的三大方向,包括Quick(快)能力.移动端能力和集成能力.针对其中的产品"报表查看打开慢""报表开发数据同步慢&qu ...

耗时 3 天，上亿数据如何做到秒级查询？