optimize complex sql query - sql-server

Question

I am using azure sql server database. I have written one sql query to generate reprot. Here it is:
;WITH cte AS
(
SELECT ProjectID, CreatedDateUTC, ProductID, LicenseID, BackgroundID from Project p
WHERE CAST(p.CreatedDateUTC AS DATE) >= #StartDate and CAST(p.CreatedDateUTC AS DATE) <= #EndDate
and IsBackgroundUsed = 1
and s7ImageGenerated = 1 and p.SiteCode in ('b2c' )
)
SELECT ProjectID , CreatedDateUTC,
(SELECT BackgroundName from Background b WHERE b.BackgroundID = cte.BackgroundID) AS BackgroundName,
(SELECT Name FROM Product pr WHERE pr.ProductID = cte.ProductID) AS ProductName,
Case WHEN LicenseID is null THEN 'Standard' ELSE (SELECT LicenseName from License l WHERE l.LicenseID = cte.LicenseID) END AS CLA,
(SELECT PurchaseFG from Product_background pb WHERE pb.BackgroundID = cte.BackgroundID and pb.ProductId = cte.productID) AS PurchaseFG,
(SELECT FGcode from Product pr WHERE pr.ProductID = cte.ProductID) AS ProductFGCode,
--(Select dbo.[getProjectFGCodeByBackground](cte.ProductID, cte.BackgroundID)) AS FGCode,
'' AS ERPOrderNumber,
0 AS DesignQuanity
from cte
WHERE (SELECT count(*) from Approval.OrderDetail od WHERE od.ProjectID = cte.ProjectID) = 0
Is there any way to optimize this query. Timeout issue comes. I have written this query in store procedure and calling that store procedure using linq entity framework.
Earlier i have used join but it's more slow down so tried with sub query. Worked more then one year now not working.

Answer 1

This will definitely improve the performance, especially if the table Approval.OrderDetail is large:
...WHERE not exists
(SELECT 1 from Approval.OrderDetail od WHERE od.ProjectID = cte.ProjectID)

Answer 2

Writing a sub-select for every single field is a terrible way to retrieve data, as you'll likely end up with a lot of Loop Joins which have terrible performance over large data sets.
Your original JOIN method is the way to go, but you need to ensure you have appropriate indexes on your joining columns.
You can also replace the WHERE clause, with a LEFT JOIN and IS NULL combination
LEFT JOIN Approval.OrderDetail od
ON od.ProjectID = p.ProjectID
...
AND od.ProjectID IS NULL;
or a NOT EXISTS (although that is more likely to have to SCAN a wider range of rows for each row returned by the main query).
WHERE NOT EXISTS
(SELECT 1 FROM Approval.OrderDetail od WHERE od.ProjectID = cte.ProjectID)
In either case, make sure your Project table is appropriately indexed on (IsBackgroundUsed, s7ImageGenerated, SiteCode, CreatedDate) and that all joins are appropriately indexed.
I'd also question whether you actually need to cast your CreatedDateUTC fields to DATE types?
A possible simplification could be:
SELECT
p.ProjectID,
p.CreatedDateUTC,
b.BackgroundName,
pr.Name,
IIF(p.LicenseID IS NULL, 'Standard', l.LicenseName) AS CLA,
pb.PurchaseFG,
pr.FGCode AS ProductFGCode,
'' AS ERPOrderNumber,
0 AS DesignQuantity
FROM Project p
LEFT JOIN Approval.OrderDetail od
ON od.ProjectID = p.ProjectID
LEFT JOIN Background b
ON b.BackgroundID = p.BackgroundID
LEFT JOIN Product pr
ON pr.ProductID = p.ProductID
LEFT JOIN License l
ON l.LicenseID = p.LicenseID
LEFT JOIN Product_Background pb
ON pb.BackgroundID = p.BackgroundID
AND pb.ProductID = p.ProductID
WHERE p.CreatedDateUTC >= #StartDate AND p.CreatedDateUTC <= #EndDate
AND p.IsBackgroundUsed = 1
AND p.s7ImageGenerated = 1
AND p.SiteCode = 'b2c'
AND od.ProjectID IS NULL;

Answer 3

WHERE CAST(p.CreatedDateUTC AS DATE) >= #StartDate and CAST(p.CreatedDateUTC AS DATE) <= #EndDate
make this SARGAble ,create non clustered index on CreatedDateUTC
Suppose this is the parameter ,
declare #StartDate datetime='2018-02-01'
declare #EndDate datetime='2018-02-28'
Then,
set #EndDate=dateadd(second,-1,dateadd(day,1,#EndDate))
now you can safely use do this,
WHERE p.CreatedDateUTC >= #StartDate and p.CreatedDateUTC <= #EndDate
I think,#Mark Sinkinson query will work ok than sub query.( I will try NOT EXISTS clause once)
Use INNER JOIN if possible.
Hope you are using Store Procedure and calling the SP.
Create index on all joins columns.
Since your sub query is working fine output wise without TOP 1 so it appear that all tables have ONE to ONE relation with Project .
CREATE NONCLUSTERED INDEX IX_Project ON project (
CreatedDateUTC
,IsBackgroundUsed
,s7ImageGenerated
,SiteCode
) include (ProductID,LicenseID,BackgroundID);
Hope projectID is already Clustered Index.

Answer 4

Might not be much faster but easier to read for me.
You should be able to adjust #StartDate and #EndDate and not have to cast to date.
Have an index on all join and where conditions.
If those are FK you should be able to use an inner join (and should).
SELECT P.ProjectID , P.CreatedDateUTC,
b.BackgroundName,
pr.Name AS ProductName,
isnull(l.LicenseName, 'Standard') as CLA,
pb.PurchaseFG,
pr.FGcode AS ProductFGCode,
'' AS ERPOrderNumber,
0 AS DesignQuanity
from Project p
left join Background b
on b.BackgroundID = p.BackgroundID
left join Product pr
on pr.ProductID = p.ProductID
left join License l
on l.LicenseID = p.LicenseID
left join Product_background pb
on pb.BackgroundID = p.BackgroundID
and pb.ProductId = p.productID
left join Product pr
on pr.ProductID = p.ProductID
WHERE CAST(p.CreatedDateUTC AS DATE) >= #StartDate
and CAST(p.CreatedDateUTC AS DATE) <= #EndDate
and p.IsBackgroundUsed = 1
and p.s7ImageGenerated = 1
and p.SiteCode = 'b2c'
and not exists (SELECT 1
from Approval.OrderDetail od
WHERE od.ProjectID = p.ProjectID)