Basic Update Question

Question

Dear Group,
I have two tables in two different R2 databases. We will call them Table1 and Table2 for convenience. I would like to update records in Table2 from the values of rows in Table1.
 I would like to loop through Table1 and find the corresponding record in Table2 and update the fields Table2.PriceChangeDate (datetime) – ‘2014-04-01 00:00:00.000’ and Table2.PriceChange (float) with the value of the Table1.Price (float).
So to me it would be:
Update Table2.PriceChangeDate = ‘2014-04-01 00:00:00.000’ AND Table2.PriceChange
 = Table1.Price
WHERE
Table2.ProductID  = Table1.ProductID
 AND Table2.PriceLevelID = Table1.PriceLevelID  AND Table2.TimeBand
 = Table2.TimeBand
The table structure is:
[Database1].[Table1]:
ID (nvarchar(15)) - Identity
ProductID (nvarchar(15))
PriceLevelID (nvarchar(15))
TimeBand (nvarchar(15))
Price (float)
[Database2].[Table2]:
ID (nvarchar(15)) - Identity
ProductID (nvarchar(15))
PriceLevelID (nvarchar(15))
TimeBand (nvarchar(15))
PriceChangeDate (datetime)
PriceChange (float)
Thanks very much for your help. It will be easy for the group I am sure but it has me slightly confused.
Alastair (from UK)

Answer 1

The one advantage with using MERGE over UPDATE is that the join conditions you causes ambiguity, this will result in error, whereas with UPDATE you will get an indeterministic result.
The potential ambiguity here is that if one row in the target table matches several rows in the source, this raises the question which price you want to copy. This will not happen if you join on a complete key in the source table, so it may not be a concern
at all.
The other advantage with MERGE over this specific form of UPDATE is that MERGE is ANSI-compliant syntax, whereas this form of UPDATE is not. There is another form of UPDATE which is both ANSI-compliant, and which also will raise an error in case of ambiguities:
Update DB1.dbo.Table1
SET  PriceChangeDate = '2014-04-01T00:00:00.000'
    ,PriceChange     =
       (SELECT t1.Price
        FROM   DB2.dbo.Table2 AS t2
        WHERE  DB1.dbo.Table1.ProductID = t2.ProductID
          AND  DB1.dbo.Table1.PriceLevelID = t2.PriceLevelID
          AND  DB1.dbo.Table1.TimeBand  = t2.TimeBand)
WHERE EXISTS (SELECT *
        FROM   DB2.dbo.Table2 AS t2
        WHERE  DB1.dbo.Table1.ProductID = t2.ProductID
          AND  DB1.dbo.Table1.PriceLevelID = t2.PriceLevelID
          AND  DB1.dbo.Table1.TimeBand  = t2.TimeBand)
But as you see, this model is even more cumbersome than the MERGE statement, and many unexperienced programmers lose themselves in this syntax.
In practice, most people in the SQL Server world use the syntax that Dan used, and as long as you aware of the risks it incurs, there is no problem with it.
Erland Sommarskog, SQL Server MVP, [email protected]