Link to home
Start Free TrialLog in
Avatar of sas77
sas77

asked on

T SQL LOOKUP QUERY

Dear Experts,


       I am loading a table from a source which is also a sql server 2008 table and it has following structure with sample rows.

OrderDetailID RecognitionDate         RecognizedAmount
------------- ----------------------- ---------------------
4262695       2010-04-15 00:00:00     -0.0049
4262756       2010-04-15 00:00:00     -0.0049
4263732       2010-04-15 00:00:00     -0.0049
4263793       2010-04-15 00:00:00     -0.0049
4262389       2010-04-15 00:00:00     -0.0023
4215156       2010-04-15 00:00:00     -0.0023
4215157       2010-04-15 00:00:00     -0.0023
4215217       2010-04-15 00:00:00     -0.0023
4215218       2010-04-15 00:00:00     -0.0023
4261839       2010-04-15 00:00:00     -0.0023

i want to load only  the records which is not existing in target that means it should strictly match complete row and see whether it is exisitng in the target or not.If not then load else do not load.

Note : as this is very huge table which consist of 7 million records please provide me with optimized query.

Thanks.
Avatar of momi_sabag
momi_sabag
Flag of United States of America image

insert into newTable
select *
from oldTable t1 left outer join newTable t2 on
t1.OrderDetailID = t2.OrderDetailID
and t1.RecognitionDate        = t2.RecognitionDate        
and t1.RecognizedAmount = t2.RecognizedAmount
Avatar of Chris Luttrell
The best solution if all the columns match is to use Set Logic which is what relational databases are best at.  Use the EXCEPT operator to identify the records that are not already in the table and inset them like this
INSERT INTO NewTable (OrderDetailID, RecognitionDate, RecognizedAmount)
SELECT OrderDetailID, RecognitionDate, RecognizedAmount
FROM OldTable
EXCEPT
SELECT OrderDetailID, RecognitionDate, RecognizedAmount
FROM NewTable;

Open in new window

Avatar of sas77
sas77

ASKER

Attached code which i am using to load fact table called fact revenue recognition, this is a incremental load and unfortunately i dont have any last_update date in source level where i can load only previous day records which should look on fact table if exisiting dont load if not load. Please provide me appropriate code which i can put in  OLEDB SOURCE IN ssis package.


Thanks.
Target Table:
 
CREATE TABLE [dbo].[FactRevenueRecognition](
	[RevenueRecognitionUID] [int] IDENTITY(1,1) NOT NULL,
	[OrderDetailKey] [int] NOT NULL,
	[OrderKey] [int] NOT NULL,
	[ProductKey] [int] NOT NULL,
	[AddressKey] [int] NOT NULL,
	[RecognitionDateKey] [int] NOT NULL,
	[RevenueRecognitionType] [char](2) NOT NULL,
	[RecognizedAmount] [smallmoney] NOT NULL,
 CONSTRAINT [RevenueRecognition_Fact_pk] PRIMARY KEY CLUSTERED 
(
	[RevenueRecognitionUID] ASC
)WITH (PAD_INDEX  = OFF, STATISTICS_NORECOMPUTE  = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS  = ON, ALLOW_PAGE_LOCKS  = ON) ON [PRIMARY]
) ON [PRIMARY]
 
GO
 
 
Source Query:
 
SELECT 
ISNULL(RR.ORDERDETAILID,'0') AS OrderDetailID,
ISNULL(O.OrderID,'0') AS OrderID,
ISNULL(P.PRODUCTID,'0') AS ProductID,
ISNULL(A.AddressID,'0') AS AddressID,
ISNULL(Production.dbo.TruncToDay(RR.RECOGNITIONDATE),'1990-01-01')AS RecognitionDate,
ISNULL(RRT.REVENUERECOGNITIONTYPE,'Unknown') AS RevenueRecognitionType,
ISNULL(RR.RECOGNIZEDAMOUNT,'0.00')AS RecognizedAmount
FROM
Production.dbo.REVENUERECOGNITION AS RR LEFT JOIN Production.dbo.ORDERDETAIL OD
ON RR.ORDERDETAILID = OD.ORDERDETAILID
LEFT JOIN 
Production.dbo.PRODUCT AS P ON OD.PRODUCTID = P.PRODUCTID
LEFT JOIN
Production.dbo.Orders AS O  ON OD.ORDERID = O.ORDERID
LEFT JOIN 
Production.dbo.OrderAddress AS OA ON O.ORDERID = OA.ORDERID
LEFT JOIN
Production.dbo.ADDRESS  AS A ON  OA.ADDRESSID = A.ADDRESSID
LEFT JOIN
Production.dbo.SEMINARINFO AS SI ON P.PRODUCTID = SI.PRODUCTID
LEFT JOIN 
Production.dbo.REVENUERECOGNITIONTYPE RRT  ON P.REVENUERECOGNITIONTYPE = RRT.REVENUERECOGNITIONTYPE

Open in new window

This should work logically but it would be much better if you could that last update date into your source table to avoid having to do this.
INSERT INTO dbo.FactRevenueRecognition
SELECT 
ISNULL(RR.ORDERDETAILID,'0') AS OrderDetailID,
ISNULL(O.OrderID,'0') AS OrderID,
ISNULL(P.PRODUCTID,'0') AS ProductID,
ISNULL(A.AddressID,'0') AS AddressID,
ISNULL(Production.dbo.TruncToDay(RR.RECOGNITIONDATE),'1990-01-01')AS RecognitionDate,
ISNULL(RRT.REVENUERECOGNITIONTYPE,'Unknown') AS RevenueRecognitionType,
ISNULL(RR.RECOGNIZEDAMOUNT,'0.00')AS RecognizedAmount
FROM
Production.dbo.REVENUERECOGNITION AS RR LEFT JOIN Production.dbo.ORDERDETAIL OD
ON RR.ORDERDETAILID = OD.ORDERDETAILID
LEFT JOIN 
Production.dbo.PRODUCT AS P ON OD.PRODUCTID = P.PRODUCTID
LEFT JOIN
Production.dbo.Orders AS O  ON OD.ORDERID = O.ORDERID
LEFT JOIN 
Production.dbo.OrderAddress AS OA ON O.ORDERID = OA.ORDERID
LEFT JOIN
Production.dbo.ADDRESS  AS A ON  OA.ADDRESSID = A.ADDRESSID
LEFT JOIN
Production.dbo.SEMINARINFO AS SI ON P.PRODUCTID = SI.PRODUCTID
LEFT JOIN 
Production.dbo.REVENUERECOGNITIONTYPE RRT  ON P.REVENUERECOGNITIONTYPE = RRT.REVENUERECOGNITIONTYPE
EXCEPT
SELECT 	OrderDetailKey,
	OrderKey,
	ProductKey,
	AddressKey,
	RecognitionDateKey,
	RevenueRecognitionType,
	RecognizedAmount
FROM dbo.FactRevenueRecognition

Open in new window

Avatar of sas77

ASKER

Dear CGLuttrell:

 That query is taking long time as it should look up on 7 million records can you please provide me any other way of doing this?
If RR.RECOGNITIONDATE in your Production.dbo.REVENUERECOGNITION table is always increasing (you would never insert an older date or update existing data) they you could try something like below for the selection of new values to insert in leu of having a last_updated_date column.  You have not given exact names of the RecognitionDateKey Dimension table or columns so those are my best guesses.
SELECT 
ISNULL(RR.ORDERDETAILID,'0') AS OrderDetailID,
ISNULL(O.OrderID,'0') AS OrderID,
ISNULL(P.PRODUCTID,'0') AS ProductID,
ISNULL(A.AddressID,'0') AS AddressID,
ISNULL(Production.dbo.TruncToDay(RR.RECOGNITIONDATE),'1990-01-01')AS RecognitionDate,
ISNULL(RRT.REVENUERECOGNITIONTYPE,'Unknown') AS RevenueRecognitionType,
ISNULL(RR.RECOGNIZEDAMOUNT,'0.00')AS RecognizedAmount
FROM
Production.dbo.REVENUERECOGNITION AS RR LEFT JOIN Production.dbo.ORDERDETAIL OD
ON RR.ORDERDETAILID = OD.ORDERDETAILID
LEFT JOIN 
Production.dbo.PRODUCT AS P ON OD.PRODUCTID = P.PRODUCTID
LEFT JOIN
Production.dbo.Orders AS O  ON OD.ORDERID = O.ORDERID
LEFT JOIN 
Production.dbo.OrderAddress AS OA ON O.ORDERID = OA.ORDERID
LEFT JOIN
Production.dbo.ADDRESS  AS A ON  OA.ADDRESSID = A.ADDRESSID
LEFT JOIN
Production.dbo.SEMINARINFO AS SI ON P.PRODUCTID = SI.PRODUCTID
LEFT JOIN 
Production.dbo.REVENUERECOGNITIONTYPE RRT  ON P.REVENUERECOGNITIONTYPE = RRT.REVENUERECOGNITIONTYPE
WHERE RR.RECOGNITIONDATE > (SELECT max(RecongnitionDateKeyValue) from RecognitionDateKeyDimensionTable)

Open in new window

Avatar of sas77

ASKER

Attached is the schema diagram  and the query i am using to populate fact table revenue recognition, recognized date will look up on dimdate and get datekey load as recognitiondatekey.

Please let me know if you need anything
RR-SCHEMA.pdf
ASKER CERTIFIED SOLUTION
Avatar of Chris Luttrell
Chris Luttrell
Flag of United States of America image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Avatar of sas77

ASKER

Thanks