关联的效率

 1 --A段
 2 ----------------------
 3 select so_d.*, 
 4        c_bf.FApproveState
 5   from sal_OrderList so_d
 6   left join bas_Bom bom_m 
 7      on (so_d.FMaterialKey = bom_m.FMaterialKey
 8         and so_d.FColorKey = bom_m.FColorKey)
 9   inner join com_BillInfo c_bf 
10      on (bom_m.FFormKey = c_bf.FFormKey) 
11         and (c_bf.FApproveState = 1)  
12 
13 --B段
14 ----------------------
15 select so_d.*, 
16        c_bf.FApproveState
17   from sal_OrderList so_d
18   left join bas_Bom bom_m
19   inner join com_BillInfo c_bf 
20      on (bom_m.FFormKey = c_bf.FFormKey) 
21         and (c_bf.FApproveState = 1)
22      on (so_d.FMaterialKey = bom_m.FMaterialKey
23         and so_d.FColorKey = bom_m.FColorKey)

 

两段脚本达到的效果是一样的

 

在查询分析器中显示计划的效果A段占用了70%的开销, B段占用30%开销，可见B段效率比A段高

 

 1 --A
 2 select wst.* 
 3   from com_MaterialWasteBook wst
 4     left join B_Material b_m on wst.FMaterialKey = b_m.FKey
 5     left join Color b_clr on wst.FColorKey = b_clr.FKey
 6     left join bas_Items b_sz on wst.FSizeKey = b_sz.FKey
 7   where b_m.FKey is not null
 8   
 9 ---B
10 select wst.* 
11   from com_MaterialWasteBook wst
12     inner join B_Material b_m on wst.FMaterialKey = b_m.FKey
13     inner join Color b_clr on wst.FColorKey = b_clr.FKey
14     inner join bas_Items b_sz on wst.FSizeKey = b_sz.FKey
15   

 上面两段

 查询开销

A 占98%

B 占2%

但是实际执行时，在100多W的记录中A执行只需25秒左右，B需要50秒（结果执行多次)