Using GROUP BY with ROLLUP, CUBE, and GROUPING SETS

In the following example, the simple GROUP BY returns a result set to compare to the result sets of examples B through K. These examples use the GROUP BY operators with the same SELECT statement.

USE AdventureWorks2008R2;
GO
SELECT T.[Group] AS N'Region', T.CountryRegionCode AS N'Country'
    ,S.Name AS N'Store', H.SalesPersonID
    ,SUM(TotalDue) AS N'Total Sales'
FROM Sales.Customer C
    INNER JOIN Sales.Store S
        ON C.StoreID  = S.BusinessEntityID
    INNER JOIN Sales.SalesTerritory T
        ON C.TerritoryID  = T.TerritoryID 
    INNER JOIN Sales.SalesOrderHeader H
        ON C.CustomerID = H.CustomerID
WHERE T.[Group] = N'Europe'
    AND T.CountryRegionCode IN(N'DE', N'FR')
    AND H.SalesPersonID IN(287, 290, 288)
    AND SUBSTRING(S.Name,1,4)IN(N'Vers', N'Spa ')
GROUP BY T.[Group], T.CountryRegionCode, S.Name, H.SalesPersonID
ORDER BY T.[Group], T.CountryRegionCode
    ,S.Name,H.SalesPersonID;

Here is the result set.

In the following example, the ROLLUP operator returns a result set that contains the following groupings:

• Region, Country, Store, and SalesPersonID

• Region, Country, and Store

• Region, and Country

• Region

• grand total

The number of groupings that is generated by ROLLUP is the same as the number of columns in the ROLLUP list plus a grand total grouping. The number of rows in a grouping is determined by the number of unique combinations of values in the columns of the grouping.

USE AdventureWorks2008R2;
GO
SELECT T.[Group] AS N'Region', T.CountryRegionCode AS N'Country'
    ,S.Name AS N'Store', H.SalesPersonID
    ,SUM(TotalDue) AS N'Total Sales' 
FROM Sales.Customer C
    INNER JOIN Sales.Store S
        ON C.StoreID  = S.BusinessEntityID
    INNER JOIN Sales.SalesTerritory T
        ON C.TerritoryID  = T.TerritoryID 
    INNER JOIN Sales.SalesOrderHeader H
        ON C.CustomerID = H.CustomerID
WHERE T.[Group] = N'Europe'
    AND T.CountryRegionCode IN(N'DE', N'FR')
    AND H.SalesPersonID IN(287, 290, 288)
    AND SUBSTRING(S.Name,1,4)IN(N'Vers', N'Spa ')
GROUP BY ROLLUP(
    T.[Group], T.CountryRegionCode, S.Name, H.SalesPersonID)
ORDER BY T.[Group], T.CountryRegionCode, S.Name, H.SalesPersonID;

Here is the result set.

In the following example, the ROLLUP operator returns a result set that contains the following groupings:

• SalesPersonID, Store, Country, and Region

• SalesPersonID, Store, and Country

• SalesPersonID, and Store

• SalesPersonID

• grand total

The columns in the ROLLUP list are the same as those in example B, but they are in the opposite order. Columns are rolled up from right to left; therefore, the order affects the groupings. The number of rows in the result set might vary with the column order.

USE AdventureWorks2008R2;
GO
SELECT T.[Group] AS N'Region', T.CountryRegionCode AS N'Country'
    ,S.Name AS N'Store', H.SalesPersonID
    ,SUM(TotalDue) AS N'Total Sales'
FROM Sales.Customer C
    INNER JOIN Sales.Store S
        ON C.StoreID  = S.BusinessEntityID 
    INNER JOIN Sales.SalesTerritory T
        ON C.TerritoryID  = T.TerritoryID 
    INNER JOIN Sales.SalesOrderHeader H
        ON C.CustomerID = H.CustomerID
WHERE T.[Group] = N'Europe'
    AND T.CountryRegionCode IN(N'DE', N'FR')
    AND H.SalesPersonID IN(287, 290, 288)
    AND SUBSTRING(S.Name,1,4)IN(N'Vers', N'Spa ')
GROUP BY ROLLUP(
    H.SalesPersonID, S.Name, T.CountryRegionCode, T.[Group])
ORDER BY H.SalesPersonID, S.Name, T.CountryRegionCode, T.[Group];

Here is the result set.

In the following example, the cross product of the two ROLLUP operations is returned.

USE AdventureWorks2008R2;
GO
SELECT T.[Group] AS N'Region', T.CountryRegionCode AS N'Country'
    ,DATEPART(yyyy,OrderDate) AS 'Year'
    ,DATEPART(mm,OrderDate) AS 'Month'
    ,SUM(TotalDue) AS N'Total Sales'
FROM Sales.Customer C
    INNER JOIN Sales.Store S
        ON C.StoreID  = S.BusinessEntityID 
    INNER JOIN Sales.SalesTerritory T
        ON C.TerritoryID  = T.TerritoryID 
    INNER JOIN Sales.SalesOrderHeader H
        ON C.CustomerID = H.CustomerID
WHERE T.[Group] = N'Europe'
    AND T.CountryRegionCode IN(N'DE', N'FR')
    AND DATEPART(yyyy,OrderDate) = '2006'
GROUP BY 
    ROLLUP(T.[Group], T.CountryRegionCode)
    ,ROLLUP(DATEPART(yyyy,OrderDate), DATEPART(mm,OrderDate))
ORDER BY T.[Group], T.CountryRegionCode
    ,DATEPART(yyyy,OrderDate), DATEPART(mm,OrderDate);

Here is the result set.

In the following example, the CUBE operator returns a result set that has one grouping for all possible combinations of columns in the CUBE list and a grand total grouping.

USE AdventureWorks2008R2;
GO
SELECT T.[Group] AS N'Region', T.CountryRegionCode AS N'Country'
    ,S.Name AS N'Store', H.SalesPersonID
    ,SUM(TotalDue) AS N'Total Sales'
FROM Sales.Customer AS C
    INNER JOIN Sales.Store AS S
        ON C.StoreID  = S.BusinessEntityID 
    INNER JOIN Sales.SalesTerritory AS T
        ON C.TerritoryID  = T.TerritoryID 
    INNER JOIN Sales.SalesOrderHeader AS H
        ON C.CustomerID = H.CustomerID
WHERE T.[Group] = N'Europe'
    AND T.CountryRegionCode IN(N'DE', N'FR')
    AND H.SalesPersonID IN(287, 288, 290)
    AND SUBSTRING(S.Name,1,4)IN(N'Vers', N'Spa ')
GROUP BY CUBE(
    T.[Group], T.CountryRegionCode, S.Name, H.SalesPersonID)
ORDER BY T.[Group], T.CountryRegionCode, S.Name, H.SalesPersonID;

Here is the result set.

In the following example, the CUBE operator returns a result set that has one grouping for all possible combinations of columns in the CUBE list and a grand total grouping.

The operator processes the grouped columns (T.[Group], T.CountryRegionCode) and (DATEPART(yyyy,OrderDate), DATEPART(mm,OrderDate)) each as a single column.

USE AdventureWorks2008R2;
GO
SELECT T.[Group] AS N'Region', T.CountryRegionCode AS N'Country'
    ,DATEPART(yyyy,OrderDate) AS 'Year'
    ,DATEPART(mm,OrderDate) AS 'Month'
    ,SUM(TotalDue) AS N'Total Sales'
FROM Sales.Customer C
    INNER JOIN Sales.Store S
        ON C.StoreID  = S.BusinessEntityID 
    INNER JOIN Sales.SalesTerritory T
        ON C.TerritoryID  = T.TerritoryID 
    INNER JOIN Sales.SalesOrderHeader H
        ON C.CustomerID = H.CustomerID
WHERE T.[Group] = N'Europe'
    AND T.CountryRegionCode IN(N'DE', N'FR')
    AND DATEPART(yyyy,OrderDate) = '2006'
GROUP BY CUBE(
    (T.[Group], T.CountryRegionCode)
    ,(DATEPART(yyyy,OrderDate), DATEPART(mm,OrderDate)))
ORDER BY T.[Group], T.CountryRegionCode
    ,DATEPART(yyyy,OrderDate), DATEPART(mm,OrderDate);

Here is the result set.

In the following example, the GROUPING SETS operator has four groupings, one for each column in the SELECT list. The operator returns one row for each unique value in the Region, Country, Store, and SalesPersonID columns.

USE AdventureWorks2008R2;
GO
SELECT T.[Group] AS N'Region', T.CountryRegionCode AS N'Country'
    ,S.Name AS N'Store', H.SalesPersonID
    ,SUM(TotalDue) AS N'Total Sales'
FROM Sales.Customer C
    INNER JOIN Sales.Store AS S
        ON C.StoreID  = S.BusinessEntityId 
    INNER JOIN Sales.SalesTerritory AS T
        ON C.TerritoryID  = T.TerritoryID 
    INNER JOIN Sales.SalesOrderHeader AS H
        ON C.CustomerID = H.CustomerID
WHERE T.[Group] = N'Europe'
    AND T.CountryRegionCode IN(N'DE', N'FR')
    AND H.SalesPersonID IN(287, 288, 290)
    AND SUBSTRING(S.Name,1,4)IN(N'Vers', N'Spa ')
GROUP BY GROUPING SETS
    (T.[Group], T.CountryRegionCode, S.Name, H.SalesPersonID)
ORDER BY T.[Group], T.CountryRegionCode, S.Name, H.SalesPersonID;

Here is the result set.

In the following example, the GROUPING SETS list contains two composite elements, (T.[Group], T.CountryRegionCode) and (DATEPART(yyyy,OrderDate), DATEPART(mm,OrderDate)). Each composite element is treated as one column.

USE AdventureWorks2008R2;
GO
SELECT T.[Group] AS N'Region', T.CountryRegionCode AS N'Country'
    ,DATEPART(yyyy,OrderDate) AS 'Year'
    ,DATEPART(mm,OrderDate) AS 'Month'
    ,SUM(TotalDue) AS N'Total Sales'
FROM Sales.Customer C
    INNER JOIN Sales.Store AS S
        ON C.StoreID  = S.BusinessEntityID 
    INNER JOIN Sales.SalesTerritory AS T
        ON C.TerritoryID  = T.TerritoryID 
    INNER JOIN Sales.SalesOrderHeader AS H
        ON C.CustomerID = H.CustomerID
WHERE T.[Group] = N'Europe'
    AND T.CountryRegionCode IN(N'DE', N'FR')
    AND DATEPART(yyyy,OrderDate) = '2006'
GROUP BY GROUPING SETS(
    (T.[Group], T.CountryRegionCode)
    ,(DATEPART(yyyy,OrderDate), DATEPART(mm,OrderDate)))
ORDER BY T.[Group], T.CountryRegionCode
    ,DATEPART(yyyy,OrderDate), DATEPART(mm,OrderDate);

Here is the result set.

In the following example, the GROUPING SETS list has five elements. The result set has one row for the following elements:

• Each unique combination of values in the Region and Country columns

• Each unique value in the Store column

• Each unique combination of values in the SalesPersonID and Region columns

• Each unique value in the SalesPersonID column

• A grand total

USE AdventureWorks2008R2;
GO
SELECT T.[Group] AS N'Region', T.CountryRegionCode AS N'Country'
    ,S.Name AS N'Store', H.SalesPersonID
    ,SUM(TotalDue) AS N'Total Sales'
FROM Sales.Customer C
    INNER JOIN Sales.Store AS S
        ON C.StoreID  = S.BusinessEntityID 
    INNER JOIN Sales.SalesTerritory AS T
        ON C.TerritoryID  = T.TerritoryID 
    INNER JOIN Sales.SalesOrderHeader AS H
        ON C.CustomerID = H.CustomerID
WHERE T.[Group] = N'Europe'
    AND T.CountryRegionCode IN(N'DE', N'FR')
    AND H.SalesPersonID IN(287, 288, 290)
    AND SUBSTRING(S.Name,1,4)IN(N'Vers', N'Spa ')
GROUP BY GROUPING SETS(
    (T.[Group], T.CountryRegionCode)
    ,(S.Name)
    ,(H.SalesPersonID,T.[Group])
    ,(H.SalesPersonID)
    ,())
ORDER BY T.[Group], T.CountryRegionCode, S.Name, H.SalesPersonID;

Here is the result set.

In the following example, the GROUPING SETS list includes groupings for columns T.[Group] and T.CountryRegionCode and a ROLLUP of columns S.Name and H.SalesPersonID.

USE AdventureWorks2008R2;
GO
SELECT T.[Group] AS N'Region', T.CountryRegionCode AS N'Country'
    ,S.Name AS N'Store', H.SalesPersonID
    ,SUM(TotalDue) AS N'Total Sales'
FROM Sales.Customer C
    INNER JOIN Sales.Store AS S
        ON C.StoreID  = S.BusinessEntityID 
    INNER JOIN Sales.SalesTerritory AS T
        ON C.TerritoryID  = T.TerritoryID 
    INNER JOIN Sales.SalesOrderHeader AS H
        ON C.CustomerID = H.CustomerID
WHERE T.[Group] = N'Europe'
    AND T.CountryRegionCode IN(N'DE', N'FR')
    AND H.SalesPersonID IN(287, 288, 290)
    AND SUBSTRING(S.Name,1,4)IN(N'Vers', N'Spa ')
GROUP BY GROUPING SETS(
    T.[Group], T.CountryRegionCode
   ,ROLLUP(S.Name, H.SalesPersonID))
ORDER BY T.[Group], T.CountryRegionCode, S.Name, H.SalesPersonID;

Here is the result set.

In the following example, the GROUPING SETS list includes groupings for columns T.[Group] and T.CountryRegionCode and a CUBE of columns S.Name and H.SalesPersonID.

USE AdventureWorks2008R2;
GO
SELECT T.[Group] AS N'Region', T.CountryRegionCode AS N'Country'
    ,S.Name AS N'Store', H.SalesPersonID
    ,SUM(TotalDue) AS N'Total Sales'
FROM Sales.Customer C
    INNER JOIN Sales.Store AS S
        ON C.StoreID  = S.BusinessEntityID 
    INNER JOIN Sales.SalesTerritory AS T
        ON C.TerritoryID  = T.TerritoryID 
    INNER JOIN Sales.SalesOrderHeader AS H
        ON C.CustomerID = H.CustomerID
WHERE T.[Group] = N'Europe'
    AND T.CountryRegionCode IN(N'DE', N'FR')
    AND H.SalesPersonID IN(287, 288, 290)
    AND SUBSTRING(S.Name,1,4)IN(N'Vers', N'Spa ')
GROUP BY GROUPING SETS(
    T.[Group], T.CountryRegionCode
    ,CUBE(S.Name, H.SalesPersonID))
ORDER BY T.[Group], T.CountryRegionCode, S.Name, H.SalesPersonID;

Here is the result set.