Question

我不是一个SQL专家，只是知道我的方式。

我需要压缩一些数据，以便为SQL报告目的生成它。

此处我的查询我不认为是最佳的，因为它很慢，我确信有10种更好的方法可以做到这一点：

此处的核心列是InspectionID，它是Inspections表的主键

有什么建议吗？

SELECT      
    INSP.HouseSiteNo,
    (SELECT TOP 1 WORKFLOWSTATES.Name FROM INSPECTIONWORKFLOWITEM WFITEM 
        INNER JOIN WORKFLOWSTATES ON WFITEM.WorkflowStateID = WORKFLOWSTATES.WorkflowStateID
        WHERE WFITEM.InspectionID = INSP.InspectionID AND WORKFLOWSTATES.StateNumber = 1) as StateName
    FROM INSPECTIONS INSP       
ORDER BY HouseSiteNo

表格结构如下：

检查：

HouseSiteNo     HouseType    ProjectID                              InspectionID
1                A           543C6381-8704-4129-B484-05B02927F690   E061F76F-8998-450B-B08C-DA2817B310A4
2                A           543C6381-8704-4129-B484-05B02927F690   E0D1F3DB-51D4-465A-A0B4-ACFFF181DC56

INSPECTIONWORKFLOWITEM

InspectionID                            WorkflowStateID                         DateStarted               StateCompleted    ID
FAA4E366-1D57-4699-982C-001FF351D717    AB2B7DA0-A90A-46FB-8F4B-8E662191E380    2012-09-09 00:00:00.000 0   D77332CB-D026-409A-814D-5632C109850A
FAA4E366-1D57-4699-982C-001FF351D717    E2AC895A-0986-4A03-B104-A727D41405E3    2012-09-09 00:00:00.000 0   F118FD02-20C4-4B45-8539-6F8BDC6B3868
FAA4E366-1D57-4699-982C-001FF351D717    F0469593-1B6B-4F98-92AE-DF50E76B9ED8    2012-09-09 00:00:00.000 0   8FBEE61B-D7B3-4D13-9FC8-730EB0FE7C95
FAA4E366-1D57-4699-982C-001FF351D717    5929D0CA-E8B0-42EE-9CF4-E3FBC9973AA6    2012-09-09 00:00:00.000 0   02646FE5-B338-4D6F-B5F5-7984B0EF7BF5
FAA4E366-1D57-4699-982C-001FF351D717    ECDDB790-88A4-4654-889D-BCA577C35CA8    2012-09-09 00:00:00.000 0   0B9C26FA-8D9D-4552-A3FE-84D4D21F5791
FAA4E366-1D57-4699-982C-001FF351D717    099D66A4-16F2-4E72-A42A-06B4D6DC0102    2012-09-09 00:00:00.000 0   3B90F113-5194-4ED2-B920-8DA004C9EA5C
FAA4E366-1D57-4699-982C-001FF351D717    66132F53-0A0E-431A-AD84-ADE77656CDF0    2012-09-09 00:00:00.000 0   05223484-8BB3-43A8-A9A8-C8FBC6270186
9AC52A0A-BC2D-4BBD-9585-0028C07ACC27    F0469593-1B6B-4F98-92AE-DF50E76B9ED8    2012-09-09 00:00:00.000 0   A1471AF4-B733-488C-86B4-FB29970E536E
9AC52A0A-BC2D-4BBD-9585-0028C07ACC27    ECDDB790-88A4-4654-889D-BCA577C35CA8    2012-09-09 00:00:00.000 0   30F3A1F9-6E51-4504-A9FD-B32FD2B35B39
9AC52A0A-BC2D-4BBD-9585-0028C07ACC27    5929D0CA-E8B0-42EE-9CF4-E3FBC9973AA6    2012-09-09 00:00:00.000 0   760D4907-ED6C-4D84-AED9-775FBBAE5123
9AC52A0A-BC2D-4BBD-9585-0028C07ACC27    AB2B7DA0-A90A-46FB-8F4B-8E662191E380    2012-09-09 00:00:00.000 0   49D632C7-31D0-4818-AC36-564F157D1959
9AC52A0A-BC2D-4BBD-9585-0028C07ACC27    66132F53-0A0E-431A-AD84-ADE77656CDF0    2012-09-09 00:00:00.000 0   C6771081-198D-475B-BA09-14BB696567DA
9AC52A0A-BC2D-4BBD-9585-0028C07ACC27    099D66A4-16F2-4E72-A42A-06B4D6DC0102    2012-09-09 00:00:00.000 0   F265B140-25C5-46CA-B060-2A27CD9F9D00
9AC52A0A-BC2D-4BBD-9585-0028C07ACC27    E2AC895A-0986-4A03-B104-A727D41405E3    2012-09-09 00:00:00.000 0   E56AEC15-4BA1-4089-B1A7-3BF054568929

WORKFLOWSTATES

WorkflowStateID                         Name          StateNumber
099D66A4-16F2-4E72-A42A-06B4D6DC0102    State 3     3
AB2B7DA0-A90A-46FB-8F4B-8E662191E380    State 6     6
E2AC895A-0986-4A03-B104-A727D41405E3    State 1     1
66132F53-0A0E-431A-AD84-ADE77656CDF0    State 2     2
ECDDB790-88A4-4654-889D-BCA577C35CA8    State 5     5
F0469593-1B6B-4F98-92AE-DF50E76B9ED8    State 7     7
5929D0CA-E8B0-42EE-9CF4-E3FBC9973AA6    State 4     4

Answer 1

如果您使用的是SQL Server 2005或更高版本，则可以使用cross/outer apply：

SELECT  INSP.HouseSiteNo, StateName.Name
FROM INSPECTIONS INSP 
CROSS APPLY
(
    SELECT TOP 1 WORKFLOWSTATES.Name FROM INSPECTIONWORKFLOWITEM WFITEM 
    INNER JOIN WORKFLOWSTATES ON WFITEM.WorkflowStateID = WORKFLOWSTATES.WorkflowStateID
    WHERE WFITEM.InspectionID = INSP.InspectionID AND WORKFLOWSTATES.StateNumber = 1
) StateName
ORDER BY HouseSiteNo

~~当我在一个批处理中运行这两个查询时，实际执行计划显示相对于批处理的查询成本是68％而我的是32％ - 因此它获得了巨大的性能提升。~~性能提升是由于缺乏ORDER BY声明造成的。目前它们都是相对于批次的50％。

Answer 2

不确定这是否是您正在寻找的。

如果您的查询需要很长时间并且有效，请尝试在分析器中运行它，看看是否有可以添加的索引以提高性能。

SELECT i.HouseSiteNo, w.Name
FROM Inspections i
JOIN InspectionWorkflowItem wfi on i.InspectionID = wfi.InspectionID
JOIN WorkflowStatus w on w.WorkflowStateID = wfi.WorkflowStateID
WHERE w.StateNumber = 1
ORDER BY i.HouseSiteNo

Answer 3

这是一个潜在的问题。没有独特排序的前1名是非确定性的即使是具有群集PK的表也不能保证在没有排序的情况下返回相同的前1个。

SELECT TOP 1 WORKFLOWSTATES.Name 
FROM INSPECTIONWORKFLOWITEM WFITEM 
INNER JOIN WORKFLOWSTATES 
   ON WFITEM.WorkflowStateID = WORKFLOWSTATES.WorkflowStateID
WHERE WFITEM.InspectionID = INSP.InspectionID 
  AND WORKFLOWSTATES.StateNumber = 1


SELECT INSP.HouseSiteNo, min(WORKFLOWSTATES.Name)  -- at least min is deterministic 
FROM FROM INSPECTIONS INSP 
JOIN INSPECTIONWORKFLOWITEM WFITEM
  ON WFITEM.InspectionID = INSP.InspectionID
JOIN WORKFLOWSTATES 
  ON WORKFLOWSTATES.WorkflowStateID = WFITEM.WorkflowStateID
 AND WORKFLOWSTATES.StateNumber = 1
GROUP BY INSP.HouseSiteNo
ORDER BY INSP.HouseSiteNo

SQL展平数据

3 个答案: