Teradata SQL Grouping – Advanced Options

Teradata SQL and Advanced GROUPING Functions

We will explain the advanced GROUPING functions using the following example. The basis for this is a table that contains the number of flights per aircraft and date:

CREATE MULTISET TABLE Flights
(
   PLANE AS BIGINT NOT NULL,
   FLIGHTDATE AS DATE NOT NULL,
   NR_FLIGHTS INTEGER NOT NULL
) PRIMARY INDEX (PLANE);

Below you can see our test data:

Teradata SQL – GROUP BY GROUPING SETS

Groups data either on one dimension without ROLLUP or on several dimensions one CUBE. Here is an example:

SELECT 
    PLANE, 
    EXTRACT(MONTH FROM FLIGHTDATE) AS TheMonth,  
    EXTRACT(YEAR FROM FLIGHTDATE) AS TheYear, 
    SUM(NR_FLIGHTS) AS NR_FLIGHTS
FROM 
    FLIGHTS
GROUP BY GROUPING SETS (PLANE,TheMonth,TheYear)
ORDER BY 1 DESC,2 DESC,3 DESC;

In our example, GROUP BY GROUPING SETS calculates the number of flights for each plane, each month, each year.

Details about GROUP BY GROUPING SET

• All columns of GROUPING sets must reference column names. The referencing cannot be done using the position as it is possible in simple GROUP BY (GROUP BY 1,2).
• Each GROUPING SET may contain a maximum of 190 columns.
• Large objects cannot be used in a GROUPING SET (CLOB, BLOB). They must first be cast into another data type. CLOB can be cast to CHAR or VARCHAR. BLOB can be cast to BYTE or VARBYTE.

Teradata SQL – GROUP BY ROLLUP

Groups data on a single dimension with more than one detail level. Here is an example:

SELECT  
    PLANE, 
    EXTRACT(MONTH FROM FLIGHTDATE) AS TheMonth, 
    EXTRACT(YEAR FROM FLIGHTDATE) AS TheYear, 
    SUM(NR_FLIGHTS) AS NR_FLIGHTS 
 FROM 
    FLIGHTS
GROUP BY ROLLUP (Plane,TheMonth,TheYear) 
ORDER BY 1 DESC,2 DESC,3 DESC;

GROUP BY ROLLUP calculates the number of flights for each plane, each distinct month, each month per year, each year plus the total.

Details about GROUP BY ROLLUP

• Several GROUPING SETS can be defined in one SQL. In our example, we use only one (Plane, TheMonth, TheYear)
• Large objects cannot be used in a GROUPING SET (CLOB, BLOB). They must first be cast into another data type. CLOB can be cast to CHAR or VARCHAR. BLOB can be cast to BYTE or VARBYTE.
• All columns of GROUPING sets must reference column names. The referencing cannot be done using the position as in simple GROUP BY (GROUP BY 1,2).
• GROUP BY ROLLUP on a single column corresponds to a simple GROUP BY statement
• Each GROUPING SET may contain a maximum of 190 columns.

Teradata SQL – GROUP BY CUBE

Groups the data in several dimensions. Here is an example:

SELECT 
    PLANE, 
    EXTRACT(MONTH FROM FLIGHTDATE) AS TheMonth, 
    EXTRACT(YEAR FROM FLIGHTDATE) AS TheYear,
    SUM(NR_FLIGHTS) AS NR_FLIGHTS 
FROM 
    FLIGHTS
GROUP BY CUBE (Plane,TheMonth,TheYear)
ORDER BY 1 DESC,2 DESC,3 DESC;

Details about GROUP BY CUBE

• Several GROUPING SETS can be defined in one Teradata SQL. In our example, we use only one (Plane, TheMonth, TheYear)
• All columns of GROUPING sets must reference column names. The referencing cannot be done using the position as in simple GROUP BY (GROUP BY 1,3).
• Large objects cannot be used in a GROUPING SET (CLOB, BLOB). They must first be cast into another data type. CLOB can be cast to CHAR or VARCHAR. BLOB can be cast to BYTE or VARBYTE.
• Each GROUPING SET may contain a maximum of 8 columns

Teradata advanced grouping options help to achieve grouping on different levels at the same time, thus increasing performance