Sql server 在视图中运行sql查询需要更多时间
大家好 我有一个巨大的sql查询。当我把这个查询放在一个存储的进程中时，执行它需要5秒，就像运行一个查询一样，它需要4-5秒，但是当我在一个视图中运行它时，它需要5分钟 请告诉我为什么它在视图中运行得那么慢 sql查询如下：Sql server 在视图中运行sql查询需要更多时间,sql-server,sql-server-2005,Sql Server,Sql Server 2005,大家好 我有一个巨大的sql查询。当我把这个查询放在一个存储的进程中时，执行它需要5秒，就像运行一个查询一样，它需要4-5秒，但是当我在一个视图中运行它时，它需要5分钟 请告诉我为什么它在视图中运行得那么慢 sql查询如下： CREATE VIEW dbo.Client_Billing_RS AS SELECT DISTINCT TOP (100) PERCENT CLIENT.OH_Code AS CLIENT, BUYER.OH_Cod
CREATE VIEW dbo.Client_Billing_RS
AS
SELECT DISTINCT TOP (100) PERCENT
CLIENT.OH_Code AS CLIENT,
BUYER.OH_Code AS BUYER,
dbo.Client_ReturnWK(pallet.MB_PR_CLOSED_DT) AS WEEKNUM,
dbo.Client_PadString(DATEPART(MONTH, CONVERT(varchar(8), pallet.MB_PR_CLOSED_DT, 112)), '0', 2) AS MONTH,
DATEPART(YEAR, CONVERT(varchar(8), pallet.MB_PR_CLOSED_DT, 112)) AS YEAR,
shipment.JS_ActualVolume,
shipment.JS_ActualWeight,
packing.MB_PD_Units ,
orderHeader.JD_OH_Buyer,
OrgMiscServ_1.OM_CustomAttrib3,
shipment.JS_TransportMode AS MODE,
shipment.JS_UniqueConsignRef ,
consol.JK_UniqueConsignRef,
DestRefCountry.RN_Desc,
part.OP_PartNum,
part.OP_Desc AS TITLE,
CONVERT(VARCHAR(8), part.OP_CustomAttrib1, 1) AS COVER_DATE,
docketLine.WE_CustomDate1 AS ON_SALE_DATE,
docketLine.WE_CustomAttrib3 AS US_BARCODE,
part.OP_CustomAttrib3 AS UK_BARCODE,
CASE WHEN freight IS NULL THEN 0 ELSE freight END AS FREIGHT,
CASE WHEN war IS NULL THEN 0 ELSE war END AS WAR,
CASE WHEN fuel IS NULL THEN 0 ELSE FUEL END AS FUEL,
shipment.JS_ActualChargeable * 1000 AS CHRG_KG,
shipment.JS_PackingMode,
'' AS MIN_CHRG,
BUYER.OH_FullName AS BUYER_NAME,
CASE WHEN Client_MF_Billing_Job_Cost_PIVOT.FUEL + Client_MF_Billing_Job_Cost_PIVOT.WAR IS NULL
THEN 0 ELSE Client_MF_Billing_Job_Cost_PIVOT.FUEL + Client_MF_Billing_Job_Cost_PIVOT.WAR END AS SUR_AMT1,
(packing.MB_PD_Units * part.OP_Weight) * (CASE WHEN rate IS NULL THEN 0 ELSE RATE END + (CASE WHEN JS_TransportMode = 'AIR' THEN CASE WHEN FUEL_LOOKUP IS NULL
THEN 0 ELSE FUEL_LOOKUP END ELSE 0 END)) AS TITLE_AMT,
consol.JK_CustomDate1 AS LOAD_DATE_OLD,
docket.WD_DocketID,
orderHeader.JD_IsCancelled,
CASE WHEN RATE IS NULL THEN 0 ELSE rate END AS RATE,
packing.MB_PD_Units * part.OP_Weight * CASE WHEN Client_Tariff_Job_Rate.RATE IS NULL THEN 0 ELSE rate END AS FRT_AMT,
part.OP_PK,
CASE WHEN LEN(part.OP_CustomAttrib2)= 1 THEN '0' + OP_CustomAttrib2 ELSE OP_CustomAttrib2 END AS ISSUE_NUMBERx,
'9' + LEFT(dbo.Client_PadString(part.OP_CustomAttrib2, '0', 2), 2) AS ISSUE_NUMBER,
Client_Consol.ETD AS LOAD_DATE,
CASE WHEN docketLine.WE_CustomAttrib3 IS NULL THEN '' ELSE SUBSTRING(docketLine.WE_CustomAttrib3, 7, 5) END AS UPC_CODE,
BUYER.OH_PK AS BUYER_PK,
CLIENT.OH_PK AS CLIENT_PK,
CASE WHEN LEN(SUBSTRING(part.OP_PartNum,1, CASE WHEN CHARINDEX('-', part.OP_PartNum) > 0 THEN CHARINDEX('-', part.OP_PartNum) - 1 ELSE 0 END))
= 0 THEN OU_LocalPartNumber ELSE SUBSTRING(part.OP_PartNum, 1, CASE WHEN CHARINDEX('-',
part.OP_PartNum) > 0 THEN CHARINDEX('-', part.OP_PartNum) - 1 ELSE 0 END) END AS MAG_CODE,
CASE WHEN JS_TransportMode = 'AIR' THEN (FUEL_LOOKUP) * (ROUND(packing.MB_PD_Units * part.OP_Weight + .5, 0)) ELSE 0 END AS SUR_AMT,
CASE WHEN JS_TransportMode = 'AIR' THEN FUEL_LOOKUP ELSE 0 END AS FUEL1,
Client_Tariff_Job_Rate.WAR_LOOKUP AS SUGGESTED_WAR,
Client_Tariff_Job_Rate.FUEL_LOOKUP AS SUGGESTED_FUEL,
Client_Tariff_Job_Rate.SHIPPING_LINE,
OrgMiscServ.OM_CustomAttrib1,
OrgMiscServ.OM_CustomDate1,
CLIENT.OH_PK AS ClientPK,
part.OP_Weight,
packing.MB_PD_Units * part.OP_Weight AS FRT_WEIGHT,
packing.MB_PD_Units * part.OP_Weight + Client_CHRG_PALLET.PALLET_KG_PORTION AS FRT_WEIGHT_GROSS,
(packing.MB_PD_Units * part.OP_Weight + Client_CHRG_PALLET.PALLET_KG_PORTION)
* CASE WHEN Client_Tariff_Job_Rate.RATE IS NULL THEN 0 ELSE rate END AS FRT_AMT_GROSS,
(packing.MB_PD_Units * part.OP_Weight + Client_CHRG_PALLET.PALLET_KG_PORTION) * (CASE WHEN rate IS NULL
THEN 0 ELSE RATE END + (CASE WHEN JS_TransportMode = 'AIR' THEN CASE WHEN FUEL_LOOKUP IS NULL
THEN 0 ELSE FUEL_LOOKUP END ELSE 0 END)) AS TITLE_AMT_GROSS,
orderHeader.JD_PK,
orderHeader.JD_OrderNumber,
Client_Tariff_Job_Rate.ONWARD_DELIVERY,
packing.MB_PD_Units * part.OP_Weight * CASE WHEN Client_Tariff_Job_Rate.ONWARD_DELIVERY IS NULL
THEN 0 ELSE ONWARD_DELIVERY END AS ONWARD_DELIVERY_AMT,
docketLine.WE_CustomDecimal4 AS COVER_PRICE,
CLIENT.OH_FullName,
(packing.MB_PD_Units * (part.OP_Weight + 0.009) + Client_CHRG_PALLET.PALLET_KG_PORTION)
* CASE WHEN Client_Tariff_Job_Rate.RATE IS NULL THEN 0 ELSE rate END AS FRT_AMT_GROSS_UPLIFT,
(packing.MB_PD_Units * (part.OP_Weight + 0.009) + Client_CHRG_PALLET.PALLET_KG_PORTION)
* (CASE WHEN rate IS NULL THEN 0 ELSE RATE END + (CASE WHEN JS_TransportMode = 'AIR' THEN CASE WHEN FUEL_LOOKUP IS NULL
THEN 0 ELSE FUEL_LOOKUP END ELSE 0 END)) AS TITLE_AMT_GROSS_UPLIFT,
packing.MB_PD_Units * (part.OP_Weight + 0.009) AS FRT_WEIGHT_UPLIFT,
packing.MB_PD_Units * (part.OP_Weight + 0.009) + Client_CHRG_PALLET.PALLET_KG_PORTION AS FRT_WEIGHT_GROSS_UPLIFT,
part.OP_Weight + 0.009 AS COPY_KG_UPLIFT,
(packing.MB_PD_Units * (part.OP_Weight + 0.009)) * (CASE WHEN rate IS NULL
THEN 0 ELSE RATE END + (CASE WHEN JS_TransportMode = 'AIR' THEN CASE WHEN FUEL_LOOKUP IS NULL
THEN 0 ELSE FUEL_LOOKUP END ELSE 0 END)) AS TITLE_AMT_UPLIFT,
packing.MB_PD_Units * (part.OP_Weight + 0.009) * CASE WHEN Client_Tariff_Job_Rate.RATE IS NULL THEN 0 ELSE rate END AS FRT_AMT_UPLIFT,
packing.MB_PD_Units * (part.OP_Weight + 0.009) * CASE WHEN Client_Tariff_Job_Rate.ONWARD_DELIVERY IS NULL THEN 0 ELSE ONWARD_DELIVERY END AS ONWARD_DELIVERY_AMT_UPLIFT,
packing.MB_PR_PalletRef
FROM Client_whspalletpacking packing WITH (NOLOCK)
inner join whsDocketLine docketLine WITH (NOLOCK) on packing.we_fk =docketline.we_pk
inner join whsDocket docket WITH (NOLOCK) on docketline.we_wd =docket.wd_pk
inner join client_whspallet pallet WITH (NOLOCK) on packing.MB_PD_PR = pallet.MB_PR_PK
inner join jobshipment shipment WITH (NOLOCK) on packing.mb_js= shipment.js_pk
inner join jobOrderHeader orderHeader WITH (NOLOCK) on docket.WD_ExternalReference =OrderHeader.JD_OrderNumber
inner join orgheader Client WITH (NOLOCK) on docket.wd_oh_client= Client.oh_pk
inner join orgheader Buyer WITH (NOLOCK) on packing.MB_PR_OH=Buyer.oh_pk
inner join jobconsol consol WITH (NOLOCK) on packing.mb_jk=consol.jk_pk
INNER JOIN OrgSupplierPart part WITH (NOLOCK) ON docketline.WE_OP = part.OP_PK
inner join OrgPartRelation WITH (NOLOCK) on part.op_pk=OrgPartRelation.OU_OP
LEFT JOIN RefUNLOCO As DestUNLOCO WITH (NOLOCK) On DestUNLOCO.RL_Code = shipment.JS_RL_NKDestination
LEFT JOIN RefCountry As DestRefCountry WITH (NOLOCK) On DestRefCountry.RN_PK = DestUNLOCO.RL_RN
LEFT OUTER JOIN Client_Tariff_Job_Rate WITH (NOLOCK) ON orderHeader.JD_PK = Client_Tariff_Job_Rate.JOB_ORDER_PK
LEFT OUTER JOIN Client_MF_Billing_Job_Cost_PIVOT WITH (NOLOCK) ON packing.MB_JH = Client_MF_Billing_Job_Cost_PIVOT.JR_JH
inner join Client_CHRG_PALLET WITH (NOLOCK) on packing.WE_FK = Client_CHRG_PALLET.WE_FK and packing.mb_pr_palletref=Client_CHRG_PALLET.mb_pr_palletref ---change
left outer join OrgMiscServ WITH (NOLOCK) on CLIENT.OH_PK =OrgMiscServ.OM_OH
inner join OrgMiscServ AS OrgMiscServ_1 WITH (NOLOCK) on BUYER.OH_PK =OrgMiscServ_1.OM_OH
left outer join Client_Consol WITH (NOLOCK) on consol.jk_pk=Client_Consol.CONSOL_PK
where (DestRefCountry.RN_Desc = 'United States')
ORDER BY CLIENT, TITLE, BUYER, MONTH DESC, WEEKNUM DESC, MODE DESC
根据你的评论，我怀疑你找到了这个短语 观察。。。左侧连接的WHERE将此更改为连接
LEFT JOIN RefCountry As DestRefCountry...
....
where (DestRefCountry.RN_Desc = 'United States')
根据你的评论，我怀疑你找到了这个短语 观察。。。左侧连接的WHERE将此更改为连接
LEFT JOIN RefCountry As DestRefCountry...
....
where (DestRefCountry.RN_Desc = 'United States')
CASE WHEN freight IS NULL THEN 0 ELSE freight END AS FREIGHT,
CASE WHEN freight IS NULL THEN 0 ELSE freight END AS FREIGHT,
可能在使用视图时，必须从头开始创建执行计划 我同意Vash的观点，作为vaiew运行的额外时间可能是由于编译执行计划的额外时间 试着运行这个
Set Statistics Time On
Select * from view
and then Set Statistics On
Exec yourSPHere
如果SQL Server解析和编译时间：解释了两次之间的差异，您的答案是，存储过程使用缓存的执行计划时，视图每次都必须创建一个执行计划。我同意Vash的观点，作为vaiew运行的额外时间可能是由于编译执行计划 试着运行这个
Set Statistics Time On
Select * from view
and then Set Statistics On
Exec yourSPHere
如果SQL Server解析和编译时：解释了两次之间的差异，您的答案是，存储过程使用缓存的执行计划时，视图每次都必须创建一个执行计划。您是否可以向我们展示有问题的代码？除非有更多内容，否则不应该这样做。你能提供代码吗？你在视图上有没有不在其内部的过滤器=？我使用的视图的执行查询是：选择*from Client\u Billing\r，其中weeknum='21'和year='2010'可能与你的问题无关，但排在前100%。。。订购人。。。在一个视图中，它充其量是多余的。它根本不能保证返回的行的顺序。在这种情况下，ORDER BY only用于确定TOP的组成部分。您是否可以向我们展示有问题的代码？除非有更多内容，否则它不应该这样做。你能提供代码吗？你在视图上有没有不在其内部的过滤器=？我使用的视图的执行查询是：选择*from Client\u Billing\r，其中weeknum='21'和year='2010'可能与你的问题无关，但排在前100%。。。订购人。。。在一个视图中，它充其量是多余的。它根本不能保证返回的行的顺序。在这种情况下，ORDER BY only用于确定什么构成回复的TOP感谢，但是它是否每次都从头开始创建执行计划？@Martin-当运行视图中的查询并执行SELECT*from sys.dm_tran_locks，其中resource_database_id='28'并排出现时，我在result@Amit-通常为否。该定义的长度为8081个字符。我似乎记得，对于计划缓存，很长的定义可能会得到不同的处理，但我记不清任何细节。我怀疑这能解释5分钟的差异。@Amit。您需要比较两次运行的实际执行计划，并查看谓词是否在以后应用于视图版本，如gbn的回答所示。如果是这样的话，使用参数化的内联TVF就可以很容易地解决这个问题
回复，但它是否每次都从头开始创建执行计划？@Martin-当在视图中运行查询并从sys.dm_tran_locks执行SELECT*from sys.dm_tran_locks，其中resource_database_id='28'并排出现时，我在result@Amit-通常为否。该定义的长度为8081个字符。我似乎记得，对于计划缓存，很长的定义可能会得到不同的处理，但我记不清任何细节。我怀疑这能解释5分钟的差异。@Amit。您需要比较两次运行的实际执行计划，并查看谓词是否在以后应用于视图版本，如gbn的回答所示。如果是这样的话，使用参数化的内联TVF就可以很容易地解决这个问题？很抱歉，上次的编辑毫无意义。视图无法参数化。@马丁·史密斯：大约3-4分钟，但结构不太复杂。@Vash-好的-我从未经历过这样的事情。但我也不认为我曾经像OP的帖子中那样做过20个表的连接！我想我们得等OP报告回来了。你所见过的编译时间最长的是什么？很抱歉，上次的编辑毫无意义。视图无法参数化。@马丁·史密斯：大约3-4分钟，但结构不太复杂。@Vash-好的-我从未经历过这样的事情。但我也不认为我曾经像OP的帖子中那样做过20个表的连接！我想我们得等OP报告回来。我不知道这个现象有名字。但我同意——看起来过滤可能被推迟了。如果我们能看到这些计划，我们肯定会知道的。我不知道这个现象有名字。但我同意——看起来过滤可能被推迟了。如果我们能看到计划，我们肯定会知道的。