Performance issue with stored procedure [closed] - sql

Question

Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
This question does not appear to be about programming within the scope defined in the help center.
Closed 7 years ago.
Improve this question
I have a stored procedure that uses the while loop temp table and cursor by which I gets the aging balance of customer, however my SP is working fine but I have some performance concerns as its take 15 sec to produce results from a small chunk of data. I am looking for a more efficient way to do this.
Thanks in advance.
Here is my stored procedure.
CREATE TABLE #Customer_Temp (
AccountCode varchar(50),
AccountTitle varchar(50),
CurrentBalance int,
FirstBalance int,
SecondBalance int,
ThirdBalance int,
FourthBalance int,
FifthBalance int,
SixthBalance int,
SeventhBalance int,
EighthBalance int,
OpeningBalance int
)
INSERT INTO #customer_temp (AccountCode, AccountTitle, OpeningBalance)
SELECT
Customer.AccountCode,
Customer.Name,
COA.OpeningBalance
FROM Customers AS Customer
INNER JOIN ChartOfAccount AS COA
ON COA.CompanyId = #Companyid
AND COA.BusinessUnitId = #BusinessUnitId
AND COA.ChartAccount = Customer.AccountCode
--Create Table And Duplicate Customers Data In it ENDED
DECLARE #DrAmount AS int
DECLARE #CrAmount AS int
DECLARE #Balance AS int
DECLARE #FBalance AS int
DECLARE #SBalance AS int
DECLARE #TBalance AS int
DECLARE #FoBalance AS int
DECLARE #FIBalance AS int
DECLARE #SIBalance AS int
DECLARE #SEBalance AS int
DECLARE #EBalance AS int
DECLARE #FSDate AS date
DECLARE #FLDate AS date
DECLARE #SSDate AS date
DECLARE #SLDate AS date
DECLARE #TSDate AS date
DECLARE #TLDate AS date
DECLARE #FOSDate AS date
DECLARE #FOLDate AS date
DECLARE #FISDate AS date
DECLARE #FILDate AS date
DECLARE #SISDate AS date
DECLARE #SILDate AS date
DECLARE #SESDate AS date
DECLARE #SELDate AS date
DECLARE #ESDate AS date
SET #FSDate = DATEADD(DAY, -1, #StartDate)
SET #FLDate = DATEADD(DAY, -6, #FSDate)
SET #SSDate = DATEADD(DAY, -1, #FLDate)
SET #SLDate = DATEADD(DAY, -6, #SSDate)
SET #TSDate = DATEADD(DAY, -1, #SLDate)
SET #TLDate = DATEADD(DAY, -14, #TSDate)
SET #FOSDate = DATEADD(DAY, -1, #TLDate)
SET #FOLDate = DATEADD(DAY, -14, #FOSDate)
SET #FISDate = DATEADD(DAY, -1, #FOLDate)
SET #FILDate = DATEADD(DAY, -14, #FISDate)
SET #SISDate = DATEADD(DAY, -1, #FILDate)
SET #SILDate = DATEADD(DAY, -29, #SISDate)
SET #SESDate = DATEADD(DAY, -1, #SILDate)
SET #SELDate = DATEADD(DAY, -89, #SESDate)
SET #ESDate = DATEADD(DAY, -1, #SELDate)
DECLARE #TempCCode AS varchar(50)
DECLARE #TempOBalance AS float
DECLARE CustomerCursor CURSOR FOR
SELECT
AccountCode,
OpeningBalance
FROM #Customer_Temp
OPEN CustomerCursor
FETCH NEXT FROM CustomerCursor INTO #TempCCode, #TempOBalance
WHILE ##FETCH_STATUS = 0
BEGIN
EXEC #FBalance = GetBalanceOfAgingOnDate #BusinessUnitId,
#Companyid,
#TempCCode,
#FSDate,
#FLDate,
#Fyear
EXEC #SBalance = GetBalanceOfAgingOnDate #BusinessUnitId,
#Companyid,
#TempCCode,
#SSDate,
#SLDate,
#Fyear
EXEC #TBalance = GetBalanceOfAgingOnDate #BusinessUnitId,
#Companyid,
#TempCCode,
#TSDate,
#TLDate,
#Fyear
EXEC #FoBalance = GetBalanceOfAgingOnDate #BusinessUnitId,
#Companyid,
#TempCCode,
#FOSDate,
#FOLDate,
#Fyear
EXEC #FIBalance = GetBalanceOfAgingOnDate #BusinessUnitId,
#Companyid,
#TempCCode,
#FISDate,
#FILDate,
#Fyear
EXEC #SIBalance = GetBalanceOfAgingOnDate #BusinessUnitId,
#Companyid,
#TempCCode,
#SISDate,
#SILDate,
#Fyear
PRINT #SESDate
PRINT #SELDate
EXEC #SEBalance =
GetBalanceOfAgingOnDate #BusinessUnitId,
#Companyid,
#TempCCode,
#SESDate,
#SELDate,
#Fyear
EXEC #EBalance = GetBalanceOfAgingOnDate #BusinessUnitId,
#Companyid,
#TempCCode,
#ESDate,
#EndDate,
#Fyear
EXEC #Balance = GetBalanceOfAgingOnDate #BusinessUnitId,
#Companyid,
#TempCCode,
#StartDate,
#EndDate,
#Fyear
UPDATE #Customer_Temp
SET CurrentBalance = (#Balance + #TempOBalance),
FirstBalance = #FBalance,
SecondBalance = #SBalance,
ThirdBalance = #TBalance,
FourthBalance = #FoBalance,
FifthBalance = #FIBalance,
SixthBalance = #SIBalance,
SeventhBalance = #SEBalance,
EighthBalance = #EBalance
WHERE AccountCode = #TempCCode
FETCH NEXT FROM CustomerCursor INTO #TempCCode, #TempOBalance
END
CLOSE CustomerCursor
DEALLOCATE CustomerCursor
AND here is the called store procedure in cursor
CREATE PROCEDURE [dbo].[GetBalanceOfAgingOnDate]
#BusinessUnitId int,
#Companyid int,
#ChartAccount as varchar (50),
#StartDate as DateTime,
#EndDate as DateTime,
#Fyear as varchar(50)
AS BEGIN
Declare #DrAmount as int
Declare #CrAmount as int
Declare #Balance as int
set #DrAmount=(select sum(Dr_Amount) from AccountVocherMaster AS AM ,
AccountVocherChild AS AC Where AM.CompanyId = #Companyid AND
AM.BusinessUnitId = #BusinessUnitId AND AM.FYear = #Fyear AND
AM.VocherId = AC.VocherId AND AC.AccountCode=#ChartAccount AND
AC.CreatedOn Between #EndDate AND #StartDate);
set #CrAmount=(select sum(Cr_Amount) from AccountVocherMaster AS AM ,
AccountVocherChild AS AC Where AM.CompanyId = #Companyid AND
AM.BusinessUnitId = #BusinessUnitId AND AM.FYear = #Fyear AND
AM.VocherId = AC.VocherId AND AC.AccountCode=#ChartAccount AND
AC.CreatedOn Between #EndDate AND #StartDate);
set #Balance = #DrAmount - #CrAmount ;
return ISNULL(#Balance,0)
END

Answer 1

make cursor a local variable DECLARE #CustomerCursor CURSOR, ensure it is not dynamic and does not reflect updates to cursor's source tables SET #CustomerCursor = CURSOR FAST_FORWARD FOR, in the end update by current cursor position, avoid additional search UPDATE #Customer_Temp SET ... WHERE CURRENT OF #CustomerCursor
obtain aggregates of different columns for similar conditions by a single select select #DrAmount = sum(Dr_Amount), #CrAmount = sum(Cr_Amount)
and more of it: select #Balance = sum(Dr_Amount) - sum(Cr_Amount)
avoid ancient-style joining by comma, write inner or outer joins, put joining conditions into ON clause, put filtering into WHERE clause
avoid thinking of SQL as a regular programming language like PHP or Pascal; it is result-set oriented; try making INLINE TABLE FUNCTION instead of procedure (GetBalanceOfAgingOnDate) - this will make you able to join to that from your queries; tip: do not make it scalar-valued or table-valued function. Look at your aggregation sp and how it is used. For each row (company) you call that sp again and again, and aggregate some values inside it. Why not to GROUP BY company and BusinessUnitID, run a single select from AccountVocherMaster and join it to your #Customer_Temp? Why not to aggregate summary for different periods by a single query? Look again: you have a list of companies, an aggregation query that provides some summarized values grouped by companies... why do you still have a cursor at all? it's a single query job
if you have lots of data and filtering by dates is the only way that makes it run fast - make not a single query, make 10 queries. this anyway will be much better than 10*2*[N companies] Actually period required is well determined - max range is [#StartDate-180, #StartDate] - and it looks like a single query is a good idea again. In some cases, if your data is very big, cursor and filtering by a scalar values may still be a good change to improve performance, thus you may still want to filter by a single CompanyID - ok, but the data you aggregate is still located in fixed date range in the same two tables - AccountVocherMaster and AccountVocherChild; loop through companies, run single aggregation query inside the cursor
there is absolutely nothing smart in your GetBalanceOfAgingOnDate's sources - don't know why you keep it as a separate module
there must be a transaction around it

Answer 2

Your problem is that you do not think SQL and try to be smart - you write procedural code, which SQL Server has to execute one by one INSTEAD of working in set and letting the query optimizer figure out how to do that most efficient.
Cursors and procedures are generally slow elements. You are better to forumulate this in as few atomic SQL statements as possible, even if the statements are a page or two long. Then the query optimizer can figure out how to achieve this result most efficient.
In your case, the second SP gets called repeatedly in the cursor - there are likely ways to achieve this better. In one or a lot less calls.

Answer 3

Hopefully this can start you down the path. There is a lot in your question, so I may not (a) understand everything here and (b) address every part, but hopefully you can take it from here. I don't see the need for any procedures or loops. Give this a try, I've commented some for context:
/* Construct a temp table to store all balance dates (and respective names) */
CREATE TABLE #BalanceDates (BalanceName VARCHAR(20), StartDate DATETIME, EndDate DATETIME);
INSERT INTO BalanceDates VALUES ('FirstBalance',DATEADD(DAY,-1,#StartDate),DATEADD(DAY,-6,#StartDate));
-- And so on with remaining inserts...
/* Create a denormalized table of all balances */
SELECT
[FirstBalance],
[SecondBalance],
...
INTO
#Balances
FROM
(
SELECT
BD.BalanceName,
ISNULL(SUM(Dr_Amount) - SUM(Cr_Amount),0) AS BalanceAmount
FROM
#BalanceDates BD
LEFT JOIN
AccountVocherChild AC
ON (AC.CreatedOn BETWEEN BD.StartDate AND BD.EndDate)
LEFT JOIN
AccountVocherMaster AM
ON (AM.VocherId = AC.VocherId)
WHERE
AM.CompanyId = #Companyid AND
AM.BusinessUnitId = #BusinessUnitId AND
AM.FYear = #Fyear AND
AC.AccountCode = #ChartAccount
GROUP BY
BD.BalanceName
) data
PIVOT
(
AVG(BalanceAmount)
FOR BalanceName IN ([FirstBalance],[SecondBalance],...)
) pvt;
/* Update the table accordingly */
UPDATE tgt
SET
FirstBalance = src.FirstBalance,
SecondBalance = src.SecondBalance,
...
FROM #Customer_Temp tgt
JOIN #Balances src
ON (1 = 1);