Question

I am assuming this is the most common scenario with everyone who is working with SQL Server.

Scenario:

I have these tables tabSRC_A(id,date,data1), tabSRC_B(id,Date,data2) and tabDEST

Now my task is to get the data from tableSRC_A, tableSRC_B apply some filtering and cleanup on them and insert them into tabDEST.

I am doing this using the following code

insert into tabDest(id, Date, Data1, Data2)
   Select id, date, Data1, Data2 
   from tabSRC_A A 
   inner join tabSRC_B B on A.id = B.id and A.date = B.date
   where not exists 
             (select * from tabDest Dest 
              where Dest.id = B.id and Dest.date = B.date)

and I am updating if already exists

Is this the best solution for this operation?

The size of the tables are 10 million rows each

I was also thinking about creating a view with surrogate key and perform a check based on the id instead of checking every row using the above method

Something like this

 insert into tabDest(id, Date, Data1, Data2)
    Select id, date, Data1, Data2 
    from view_Created_From_TabA_TabB_adding_a_SurrogateKey_Kid SV
    where SV.Kid > select (max(id) from tabDest) 

I am assuming this would be much faster.

Please guide me with any suggestions you have.

(I’m using SQL Server 2000, I know its very old)

Answer 1

Have you tried a LEFT join to detect not exists?

insert into tabDest(id, Date, Data1, Data2)
   Select id, date, Data1, Data2 
   from tabSRC_A A 
   inner join tabSRC_B B on A.id = B.id and A.date = B.date
   LEFT JOIN tabDest Dest 
   ON Dest.id = B.id and Dest.date = B.date
WHERE
   Dest.id is null