Compare SQL Server Datasets with INTERSECT and EXCEPT

Problem
When joining multiple datasets you have always had the ability to use the UNION and UNION ALL operator to allow you to pull a distinct result set (union) or a complete result set (union all).  These are very helpful commands when you need to pull data from different tables and show the results as one unified distinct result set.  On the opposite side of this it would be helpful to only show a result set where both sets of data match or only where data exists in one of the tables and not the other. This could be done with using different join types, but what other options does SQL Server offer?

SolutionWith SQL Server 2005, Microsoft introduced the INTERSECT and EXCEPT operators to further extend what you could already do with the UNION and UNION ALL operators.

• INTERSECT - gives you the final result set where values in both of the tables match
• EXCEPT - gives you the final result set where data exists in the first dataset and not in the second dataset

The advantage of these commands is that it allows you to get a distinct listing across all of the columns such as the UNION and UNION ALL operators do without having to do a group by or do a comparison of every single column.  

Like the UNION and UNION ALL operators the table structures need to be consistent as well as the columns need to have compatible data types.

Let's take for example we have two tables manager and customer.  Both of these tables have somewhat the same structure such as the following columns:

• FirstName
• LastName
• AddressLine1
• City
• StateProvinceCode
• PostalCode

Manager table sample data

Customer table sample data

We want to do two queries:

1. Find the occurrences where a manager is a customer (intersect)
2. Find the occurrences where the manager is not a customer (except)

INTERSECTIf we want to find out which people exist in both the customer table and the manager table and get a distinct list back we can issue the following command:

SELECT FIRSTNAME,         LASTNAME,         ADDRESSLINE1,         CITY,         STATEPROVINCECODE,         POSTALCODE FROM   MANAGER INTERSECT   SELECT FIRSTNAME,         LASTNAME,         ADDRESSLINE1,         CITY,         STATEPROVINCECODE,         POSTALCODE FROM   CUSTOMER

Here is the result set:

To do this same thing with a regular T-SQL command we would have to write the following:

SELECT   M.FIRSTNAME,           M.LASTNAME,           M.ADDRESSLINE1,           M.CITY,           M.STATEPROVINCECODE,           M.POSTALCODE FROM     MANAGER M           INNER JOIN CUSTOMER C             ON M.FIRSTNAME = C.FIRSTNAME                AND M.LASTNAME = C.LASTNAME                AND M.ADDRESSLINE1 = C.ADDRESSLINE1                AND M.CITY = C.CITY                AND M.POSTALCODE = C.POSTALCODE GROUP BY M.FIRSTNAME,M.LASTNAME,M.ADDRESSLINE1,M.CITY,           M.STATEPROVINCECODE,M.POSTALCODE

EXCEPTIf we want to find out which people exists in the manager table, but not in the customer table and get a distinct list back we can issue the following command:

SELECT FIRSTNAME,         LASTNAME,         ADDRESSLINE1,         CITY,         STATEPROVINCECODE,         POSTALCODE FROM   MANAGER  EXCEPT  SELECT FIRSTNAME,         LASTNAME,         ADDRESSLINE1,         CITY,         STATEPROVINCECODE,         POSTALCODE FROM   CUSTOMER

Here is the result set:

To do this same thing with a regular T-SQL command we would have to write the following:

SELECT   M.FIRSTNAME,           M.LASTNAME,           M.ADDRESSLINE1,           M.CITY,           M.STATEPROVINCECODE,           M.POSTALCODE FROM     MANAGER M WHERE    NOT EXISTS (SELECT *                       FROM   CUSTOMER C                       WHERE  M.FIRSTNAME = C.FIRSTNAME                              AND M.LASTNAME = C.LASTNAME                              AND M.ADDRESSLINE1 = C.ADDRESSLINE1                              AND M.CITY = C.CITY                              AND M.POSTALCODE = C.POSTALCODE) GROUP BY M.FIRSTNAME,M.LASTNAME,M.ADDRESSLINE1,M.CITY,           M.STATEPROVINCECODE,M.POSTALCODE

From the two examples above we can see that using the EXCEPT and INTERSECT commands are much simpler to write then having to write the join or exists statements.

To take this a step further if we had a third table (or forth...) that listed sales reps and we wanted to find out which managers were customers, but not sales reps we could do the following.

SalesRep table sample data

 

SELECT FIRSTNAME,         LASTNAME,         ADDRESSLINE1,         CITY,         STATEPROVINCECODE,         POSTALCODE FROM   MANAGER INTERSECT   SELECT FIRSTNAME,         LASTNAME,         ADDRESSLINE1,         CITY,         STATEPROVINCECODE,         POSTALCODE FROM   CUSTOMER  EXCEPT  SELECT FIRSTNAME,         LASTNAME,         ADDRESSLINE1,         CITY,         STATEPROVINCECODE,         POSTALCODE FROM   SALESREP

Here is the result set:

As you can see this is pretty simple to mix and match these statements.  In addition, you could also use the UNION and UNION ALL operators to further extend your final result sets.

Source collected from MSSQLTIPS.COm