Question

在我工作的地方，数据库已被触及＆＃34;一些不再在这里工作的人。某些数据库视图运行速度非常慢。我认为这是因为它使用了许多JOINS和UNIONS。在我看来应该有一个更有效的方法来做到这一点。原谅我在这方面缺乏知识。

        SELECT  s.id                    AS 'ShipmentID',
                s.tracking_no           AS 'Tracking Number',
                c.name                  AS 'Client Name',
             ...
            many more
             ...

        FROM    shipments_archive s
                left join client c on c.id = s.client
                left join DIM_ClientCompany cc on cc.client_id = s.client
                left join company co on co.company_id = cc.company_id
                left join weekof wo on CAST(s.date_shipped AS DATE) = CAST(wo.mydate AS DATE)
                left join DIM_ResolutionCodes res on res.Resolution_Key = s.resolutions

            WHERE
                co.company_name NOT IN ('client1', 'client2')

        UNION ALL

        SELECT  s.id                    AS 'ShipmentID',
                s.tracking_no           AS 'Tracking Number',
                c.name                  AS 'Client Name',
             ...
            many more
             ...
        FROM     shipments_archive s
                 left join client c on c.id = s.client
                 left join DIM_ClientCompany cc on cc.client_id = s.client
                 left join company co on co.company_id = cc.company_id
                 left join weekof wo on CAST(s.date_shipped AS DATE) = CAST(wo.mydate AS DATE)
                 left join DIM_ResolutionCodes res on res.Resolution_Key = s.resolutions

        WHERE co.company_name IN ('Client3', 'Client4')

        UNION ALL

        SELECT  s.id                    AS 'ShipmentID',
                s.tracking_no           AS 'Tracking Number',
                c.name                  AS 'Client Name',
             ...
            many more
             ...
        FROM    shipments_archive s
                left join client c on c.id = s.client
                left join DIM_ClientCompany cc on cc.client_id = s.client
                left join company co on co.company_id = cc.company_id
                left join DIM_Pharmacy p on p.company_id = co.company_id and p.[Lookup] = s.shipper
                left join weekof wo on CAST(s.date_shipped AS DATE) = CAST(wo.mydate AS DATE)
                left join DIM_ResolutionCodes res on res.Resolution_Key = s.resolutions
        WHERE co.company_name IN ('Client5')

        UNION ALL

您是否可以使用 GOTO 或 CASE 语句。为每个客户端多次提取此数据似乎非常耗时。有超过500万条记录。任何帮助将不胜感激。

Answer 1

从索引的角度来看，正如评论所示，查看执行计划将为如何优化此查询的索引提供一些很好的见解。从TSQL的角度来看，我不喜欢在多次使用的语句中看到相同的内容。一种选择是使用表变量。它将使用更多内存但保存IO。话虽如此，我会检查UNION ALL语句，因为看起来这个查询将返回重复记录（公司名称不在1＆amp; 2中，公司名称不在3＆amp; 4中，都返回客户端5的记录）。

--This will create a table in memory and populate it from the source tables. This way, you only read the source tables once.
DECLARE @table_variable TABLE (
ShipmentID INT
,TrackingNumber INT
,etc...)

INSERT @table_variable (ShipmentID, TrackingNumber, etc...)

SELECT
    s.id AS 'ShipmentID'
    ,s.tracking_no AS 'Tracking Number'
    ,c.name AS 'Client Name'
    --...
    --many more
    --...
FROM shipments_archive s
left join client c on c.id = s.client
left join DIM_ClientCompany cc on cc.client_id = s.client
left join company co on co.company_id = cc.company_id
left join weekof wo on CAST(s.date_shipped AS DATE) = CAST(wo.mydate AS DATE)
left join DIM_ResolutionCodes res on res.Resolution_Key = s.resolutions

--You then select from the table in memory for your subsets.
SELECT
    ShipmentID
    ,TrackingNumber
    ,etc...
    --...
    --many more
    --...
FROM @table_variable
WHERE company_name NOT IN ('client1', 'client2')

UNION ALL

SELECT
    ShipmentID
    ,TrackingNumber
    ,etc...
    --...
    --many more
    --...
FROM @table_variable
WHERE company_name NOT IN ('client3', 'client4')

UNION ALL

SELECT
    ShipmentID
    ,TrackingNumber
    ,etc...
    --...
    --many more
    --...
FROM @table_variable
WHERE company_name NOT IN ('client5')

加快MSSQL JOINED查询的建议

1 个答案: