Using SET As An Alternative To Cursors For Rolling Totals
Sep 4, 2001
I had a need to update a column in a table with a calculated value based on the value of the preceeding value in the table (which in turn was based on the preceeding value, etc...).
Although this can easily solved via a cursor, I was able to solve this with a single set statement that first updates a variable (which can be part of a calculation using the 'current' row) and using the knowledge that SET statements are applied 'in order' (top-down).
Most importantly: the technique relies on my observation that the SET statement processes a table in clustered index sequence, or load sequence in a heap table.
I've included a very simple example of this technique below.
I'm curious if anyone is using this (undocumented?) technique and if my assumptions are valid (I'm using this on much larger tables with more complex calculations).
EXAMPLE:
CREATE TABLE #tmp (
num int IDENTITY(1,1) NOT NULL,
val int NULL)
-- load 100 test rows
DECLARE @x int
SET @X = 1
WHILE @X <= 100
BEGIN
INSERT #tmp (val) VALUES (NULL)
SET @X=@X+1
END
DECLARE @i int
SET @i = 1
UPDATE #tmp -- this is a -very- simple example of technique!
SET @i=num+@i, -- @i now has current totaled value + current-row int value
val=@i -- val now stores current subtotal
These are performing extremely poorly and worse compared to our older SQL Server 2000 environment. Is there anything obvious from this formula that would make it worse?
I appreciate any guidance as to what I should look for.
I'm trying to generate a cumulative total with a rolling window of 13 time periods..Previously I was able to do left outer join to the same table 13 times to add the quantity field but it appears with the migration to SQL Server 2014, that many left outer joins is not possible (query that would run in 3 mins is taking well over 15 hours now)..
I have create a simple tabular rolling forecast report (with cumulative totals) from today (October) thru the next 12 months that looks like this. It smartly works no matter when the report is generated, by starting with this CurrentMonth and moving forward by using 1,2,3,4,etc. in the dateadd: =MonthName(datepart("m",dateadd("m",1,Now())))
The report sample (formatting lost in dropping it in here):
Code Block
Close Pct October November December January February
25.% $0 $0 $26,625 $0 $0
50.% $237,500 $110,000 $262,500 $0 $0
75.% $56,250 $0 $891,075 $0 $0
90.% $1,051,830 $0 $0 $0 $0
Monthly Total $1,345,580 $110,000 $1,180,200 $0 $0
Cumulative Total $1,345,580 $1,455,580 $2,635,780 $2,635,780 $2,635,780
It is working fine....there doesn't seem to be anything wrong with it (all numbers total correctly, etc.), but it is very unelegant.....and I know there must be a better way.
In the righthand most month (which would be September 2008) column, I have a formula that produces the amount (the Monthly Total amount is the same):
I have some data grouped in a table by a certain criteria, and for each group it is computed a subtotal for the group. Of the values from each of the group, I want to create a grand total on the report by adding every subtotal from each group.
Example: ... .... Group1 Value 10 20 Sub Total 1: 30
Group2 Value 15 25 Sub Total 2: 40
Now, I would like to be able to add subtotal 1 (30) to subtotal 2 (40) and my grand total would be 70. Can I accomplish this task in SSRS?
I have been providing sales data for a few months now from a table that is set up like this:
Date WorkDay GasSales EquipmentSales
9/1/2006 1 100.00 200.00
9/4/2006 2 50.00 45.00
etc.
As can be seen, the data is daily, i.e., on the first workday of September we sold one hundred dollars in gas and two hundred dollars in equipment. On the second workday of September we sold fifty dollars in gas and forty-five dollars in equipment.
Now, however, the data I have to pull from is cumulative. So, using the last table as an example it would look like this:
Date_WorkDay_GasSales_EquipmentSales
9/1/2006 1 100.00 200.00
9/4/2006 2 150.00 245.00
etc.
To make things more complicated, the powers that be wanted this data presented in this fashion:
Total Sales:
1_2_etc.
300.00 95.00 etc.
So, I have been doing a pivot on a CRT to get the data to look like I want. The code is like this:
with SalesCTE (Month, WorkDay, [Total Sales])
as
(
SELECT
datename(month, cag.date),
cag.WorkDay AS [Work Day],
sum(cag.sales_gas + cag.sales_hgs) AS [Total Sales]
FROM CAG INNER JOIN
Branch ON CAG.[Oracle Branch] = Branch.OracleBranch
group by cag.date, cag.WorkDay
)
select * from SalesCTE
pivot
(
sum([Total Sales])
for WorkDay
in ([1],[2],[3],[4],[5],,[7],,[9],[10],[11],[12],[13],[14],[15],[16],[17],[18],[19],[20],[21],[22],[23])
) as p
So, my question is:
How do I get the data to give back daily totals instead of the cumulative amounts for each workday? If the query was a simple one, I'd do something like
select [1] as [Day 1], [2]-[1] as [Day 2], [3]-[2] as [Day 3], etc.
but the query is far from normal, with the CRT and the pivot. I can't seem to get it to work how I'd like.
Any advice/answers? Thanks in advance!!!
P.S. I don't know how to get it to quit with the freakin' smileys.... I suppose you can figure out what my code is really supposed to look like above. Needless to say, it doesn't include a devil face and a damn music note...
I have created a DTS Package that does the following: 1 Delete all data from table 1 (SQL Task) 2 Import Data from .csv file into table 1 (Data import) 3 insert records from table 1 into table 2 when they dont exist in table 2 (SQl Task)
This all works fine, but now i want a rollback function in step 1 and 2. So when Step 1 is finished and something goes wrong i want the deleted data back.
One of our engineers here by mistake deleted some very important data without any begin trans block and we need the data back very badly.. We have not closed the session as of now.. The engineer was logged in as sa to the DB.
I'm writing a query that will be calculating a number of business performance measures including the following...
Average Stock: This is calculated by taking the actual stock value of the product category per month and dividing it by 13 if we have over 12 month's history, or if a new product, then by the number of months we have had it in stock for.
Cost of Goods Sold Annualised: This is an average figure for the year of the sold landed costs for the product category over a 13 month period if we have over 12 months history. If it's a new product, then it's for the number of months we have had stock.
My question is, how do I tell the query to go back 13 months from the current month?
We have an inventory of devices we service and wish to show the total numbers of active devices per month going 12 months back. How would I go about this?
I'm performing a stored proc that has 4 inserts. I only want the inserts to complete as a batch. If one fails, I want to rollback the whole transaction. Does anyone know the syntax?? :)
I want to insure that each of my insert statements in a stored proc are rolled back if any of the inserts fail. I already have the below statement with error handling but is this correct? It seems to me that all the steps should be made part of an entire transaction so if one part fails then it all fails. Can someone help me w/ the syntax of this??
CREATE PROCEDURE Addrecords AS
--USERS INSERT INTO [Production].[dbo].[USERS]([LastName], [UserName], [EmailAddress], [Address1], [WorkPhone], [Company], [CompanyWebsite], [pword], [IsAdmin], [IsRestricted],[AdvertiserAccountID]) SELECT dbo.fn_ReplaceTags (convert (varchar (8000),Advertisername)), [AdvertiserEmail], [AdvertiserEmail],[AdvertiserAddress], [AdvertiserPhone], [AdvertiserCompany], [AdvertiserURL], [AccountNumber],'3',0, [AccountNumber] FROM production WHERE not exists (select * from users Where users.Username = temp.AdvertiserEmail) AND validAD=1 IF @@ERROR <> 0 BEGIN ROLLBACK TRAN RETURN END
--PROPERTY INSERT INTO [Production].[dbo].[Property]([ListDate],[CommunityName],[TowerName],[PhaseName],[Unit], [Address1], [City], [State], [Zip],[IsActive],[AdPrintId]) SELECT [FirstInsertDate],[PropertyBuilding],[PropertyStreetAddress],PropertyCity + ' ' + PropertyState + ' ' + PropertyZipCode as PhaseName,[PropertyUnitNumber],[PropertyStreetAddress],[PropertyCity], [PropertyState], [PropertyZipCode],'0',[AdPrintId] FROM [Production].[dbo].[Temp] WHERE AdvertiserEmail IS NOT NULL AND validAD=1 IF @@ERROR <> 0 BEGIN ROLLBACK TRAN RETURN END
trying to develop a report that shows the sum total for each month during a specified date range.
I have parameters asking for the start date and end date of the date range, the report needs to show the months in between this date range.
One of the fields in the dataset is durationminutes which I need to sum for each month then divide by 60 to show hours. I then need to show the total of each month in a bar graph (with each month along the x axis).
If anyone can shed some light on how I am going to do this it would be greatly appreciated.
Hi All, I wanna pull a 12 month rolling data and donno how to do that, can anybody help ? Problem: lets say if i run the query today i should get data between oct-3-2006 and oct-3-2007, my main colums is monthyear which is date datatype. Appreciate your help. Thanks Chinna
I'm trying to figure out how to this do with TSQL and preferablythrough joins rather than by using cursors or temp tables. I havemultiple time spans for multiple individuals that I would like torollup together where they don't actually have a gap in coverage.ID StartDate EndDateZ000001 01/01/1996 9/30/1996Z000001 10/01/1996 12/31/1998Z000001 01/01/1999 04/30/1999Z000001 01/01/2000 12/31/2000Z000001 01/01/2001 12/31/2001I would like to be able to "roll these up" in order to reflect theactual time frames so that it would like this:ID StartDate EndDateZ000001 01/01/1996 4/30/1999Z000001 01/01/2000 12/31/20014/30/1999 to 01/01/2000 is, of course, a legitimate break where therewas a date difference of more than one day.Thanks in advance for any ideas.
I have a stored procedure that calls another stored procedure with thefirst stored procedure opening a transaction:BEGINSET XACT_ABORT ONBEGIN TRANSACTIONdoes various updates/insertscalls 2nd stored procedure to proccess updates/inserts common to manyother stored proceduresdoes more various updates/insertscommitENDThe problem I'm having is that within the 2nd stored procedure is thatif it encounters an error, it does not roll back the entiretransaction and I finish up with missing records in the database. Amusing this in the 2nd stored procedure:if(@TypeId1 = @TypeId2 and @Line1 <'' and @Line2 <'')beginRAISERROR('error message', 16, 1)RETURNendWhat could the problem be? From what I've read, it seems as thoughyou can't have an open transaction within one sp that calls another spand it maintains the same transactoin? Is this corrrect?I tired the following too, and I still couldn't get it to work. Anyideas anyone?************ sp 1 ***********Declare @AddressError char(3)SET XACT_ABORT ONBEGIN TRANSACTIONexec Sp2@AddressError OUTPUT,@variable1,@variable2,etc. etc************** sp 2 *****************@AddressError char(3) OUTPUT,if(@TypeId1 = @TypeId2 and @Line1 <'' and @Line2 <'')beginRAISERROR('error message', 16, 1)RETURNendSET XACT_ABORT ONBEGIN TRANSACTIONprocess updates/insertsSet @AddressError = 'no'Commit******** back to sp 1************If @AddressError <'no'BEGINrollback transactionENDcontinue doing updates/insertscommit
Is database mirroring compatible between SQL Server 2005 (Standard Edition) SP1 and SP2? I currently run SP1, and will shortly be introducing a failover partner. I'd like to have the failover partner installed as SP2, start the mirroring, manually failover, then upgrade the existing server to SP2 as well.
Is this a supported upgrade method, or will I need to upgrade the existing server to SP2 before setting up database mirroring? The SP2 installation instructions and readme file don't appear to mention mirroring at all (at least not that I can find).
I'm trying to create an accumulating field based on a set of records. I need to fill in daily amount balances that accumulates on a daily basis. But I can't seem to figure out how to create a total for the daily dates and have it add on additional amounts if needed.
Now I already have a table with the dates created via a stored procedure. I have a set of dates from 5/5/2000 to 5/8/2000. So that results set should look like this:
I'm trying to creating a rolling sum that accumulates the amount field for each daily record and if a new amount is listed, then roll that amount into the total. If you have any suggestions about how to perform this rolling total via TSQL or SSIS, I would greatly appreciate it.
I have a case where I read from SQL Server DB and write to a flat file.
I have one Data Flow Task inside which I have a OLEDB source component that feeds rows to a script component that writes to a flat file. I have set the txn attributes for the container to "Required" and "Read committed" . But I find that rows are written to flat file even when I throw an exception from my script component. Question is how do I prevent rows from being written to the flat file if error/exception happens. I want the whole process to be in a single transaction.
Dear All I have the following table structure and following data create table emp(empid char(4),empname char(10),empsal money,empsalPercent money,empsalRuningTot money) empid empname empsal empsalpercente001 tom 20.00 8.69e002 *** 40.00 17.39e003 harry 15.00 6.52e004 masum 25.00 10.86e005 rana 30.00 13.04e006 dipak 60.00 26.08e007 rana 40.00 17.39 now would like to make the following 1) want to sort first emp table based on empsalPercent field on descending order physicallyProb : am only able to select the same thru descending order but am unable to make it physical shortingselect * from emp order by empsalPercent desc 2) after make descending order for empsalpercent I want to make RunningTotal for the EmpsalPercent column pls helpsamir
We have installed SQL Server 2000 Service Pack 4 recently and have had various issues with some of our Stored Procedures. (which we didn't have with SP 3)
We are looking at our options regarding rolling back to Service Pack 3.
Can this be done ? & If so, is there a method documented on how to do this ?
Ive got an insert statement that fails, and below that I have code like the following:
IF @@ERROR <> 0 BEGIN -- Roll back the transaction ROLLBACK -- Raise an error and return RAISERROR ('Error INSERT INTO Address.', 16, 1) print 'test was here' RETURN END
However, there is now rollback and the inserts below it are going through.
Goal: set security on running specific stored procedures based upon user login and databse access
I have some DBA's who want to retain full control of databses / stored procedures as they now have but I want to restrict or rollback some of the changes that were implemented when the sql 2005 was set up. The sql 2005 EE is in a clustered system and uses Mixed Mode Authentication.
An example of what I want to restrict: The DBA's want to be able to view and kill processes for the different databases that are installed under their instance. The problem is other customer databases are also under the same instance.
Is their a way I can combine or have the stored procedure sp_lock only show the processes for the databases they have access to based upon their login? My concern is they will kill a process and affect the other customers.
Hi I need to write something in T-SQL which will return a record set from a table of IT hardware. The record set should comprise all items that are older than 3 years from the current date. Each hardware item has a purchase date against it so it should not be too difficult. I am considering the DATEDIFF function but I am not sure how best to write it.
Thanks for taking the time to read my post. I greatly appreciate it!
What i'm trying to do is get a distinct count of account numbers within a rolling period. My actual take is rather large but i've created a smaller-like version below. Please reference this table.
Account Date
1 1/1/08
2 1/2/08
3 1/2/08
2 2/8/08
4 2/9/08
1 2/15/08
1 3/5/08
5 3/6/08
4 3/9/08
3 3/10/08
1 4/1/08
5 4/9/08
2 4/15/08
3 4/26/08
1 5/3/08
2 5/15/08
3 5/29/08
6 5/30/08
Let's say i want to return distinct count of accounts within a 2-month rolling period meaning in February, i'd get a distinct count for accounts in January & February, then in March i'd get a distinct count for February & March, then in April i'd get it for March & April, and so on... my results table would like the table below:
Account Month
3 1
4 2
5 3
5 4
5 5
I had asked this before but it was a summing equation and not a unique count. I've tried to play with the summing equation to kind of make it work, but i'm starting to get a headache. It's probably so simple!
Here's my previous post: http://forums.microsoft.com/MSDN/ShowPost.aspx?PostID=2289509
hi, i want to create a disaster recovery site, to which i can fail over (not automatically), and also to have the option to return the database to a point in time? for example, if my principal server fails in 17:00, i want to have the option to make the mirror server available for users from 17:00 (or at least close to that time), and also to be able the return to the data from 16:00 (in the mirror site). Is it possible, and what is the best way to do it?
I would appreciate a bit of advice here. There is a largish complaint here regarding a cutomer who has entered data in 1 of our online forms, but we suspect this was then overwritten by a cached form she also had open. Anyway to cut a long story short i need to roll back the database to a point in time.
Not something i have ever had to do.
The row where the id field is equal to 3352, and this would have been written to the database at 14:58:36 on 08-aug-2005. This was over written by the data in row with id 3380 at about 11am this morning, now is this is a live database on a webserver so i cant compromise its uptime as it get written to about 3 times a minute, so how should i go about this?
I need calculating a rolling 3 month average cost from the two dataset below. Which is the 3 month Average of Dataset1 / Dataset 2.
Dataset 1:
SELECT(factAdmissions.ContractCode + '-' +factAdmissions.BenefitPlanCode) AS [Contract Code], factAdmissions.AdmitCCYYMM, ISNULL(sum(AmountPaid),0)As [Amount Paid] FROM factAdmissions
[Code] ....
Dataset2:
Select
(factMembership.ContractCode+'-'+ factMembership.BenefitPlanCode) As Product, EffectiveCCYYMM, ISNULL(count(Distinct MemberId),0) As MemberCount From factMembership Where EffectiveCCYYMM >= '200701'