So what?,

Update D Set a= S.a, b = S.b, C = S.c

From table_to D join table_from S on D.a =S.a

insert into table_to

  (A, B, C)

Select

  a, b, c

From

  table_from...

And probaly you don't even need the IsNull checking that PW wrote

insert into table_to

  (A, B, C)

Select

  a, b, c

From

  table_from

OR:

SELECT SUM(NUM) AS Total

FROM(

SELECT COUNT(*) AS NUM

FROM table1

WHERE EmpTermDate IS NULL

UNION all

SELECT COUNT(*) AS NUM

FROM table2

WHERE EmpTermDate IS NULL )

OR

SELECT (SELECT COUNT(*) AS NUM

FROM...

I am positive that 4300 IDs are FAR from the limit on an IN clause you must look else where in the process!

HTH

You are being very vage on description but I would usually say The First!

>>Be nice to developers! Sometimes I am one<<

Me TOO

>> Also, it'll ensure developers are explicitly declaring the owner of said objects, thereby eliminating a performance concern<<

And this one is another Herculian task

Right Click on task manager see how many sqlservr.exe (image name) process you see?

Have a look at "Linked Servers" on BOL

BTW the statement should use 4 part names like:

<ServerName.DBName.dbo.table>

Try:

UPDATE TEST

SET Work_Phone = P.Phone

FROM MEMBERSHIP_MTEST TEST

INNER JOIN

MEMBERSHIP AM

ON AM.ID = TEST.ID

INNER JOIN Phone P

ON AM.CUSTOMER = P.Customer

WHERE P.Phone_Type ='Off1'

In your previous you were making a CROSS JOIN of ...

At that point you will have to decide between performance and security.

If Security goes first then by all means you must use an API/Package/etc and go for it I am...

CHARINDEX(field, ‘~’ + UserID + ‘~’) > 0

That technique is sssssssssslllllllllllooooow. you can't use indexes with it. I works on small sets but you are going to get table scans....