HAVING clause

<< Click to Display Table of Contents >>

Navigation:  Streaming SQL Reference Guide > SELECT statement  >

HAVING clause

Previous pageReturn to chapter overviewNext page

The HAVING clause in a SELECT specifies a condition to apply within a group or aggregate. In other words, HAVING filters rows after the aggregation of the GROUP BY clause has been applied. Since HAVING is evaluated after GROUP BY, it can only reference expressions constructed (or derivable) from grouping keys, aggregate expressions, and constants. (These are the same rules that apply to expressions in the SELECT clause of a GROUP BY query.) A HAVING clause must come after any GROUP BY clause and before any ORDER BY clause. HAVING is like WHERE, but applies to groups. Results from a HAVING clause represent groupings or aggregations of original rows, whereas results from a WHERE clause are individual original rows.

In non-streaming applications, if there is no GROUP BY clause, GROUP BY () is assumed (though since there are no grouping expressions, expressions can consist only of constants and aggregate expressions). In streaming queries, HAVING cannot be used without a GROUP BY clause.

WHERE and HAVING can both appear in a single SELECT statement. The WHERE selects from the stream or table those individual rows that satisfy its condition (the WHERE-condition). The GROUP BY criteria apply only to the rows selected by the WHERE condition.

Such a grouping, for example "GROUP BY CustomerID", can be further qualified by a HAVING-condition, which then selects aggregations of rows satisfying its condition within the specified grouping. For example, "GROUP BY ClientID HAVING SUM(ShipmentValue) > 3600" would select only those clients whose various shipments that fit the WHERE criteria also had values that added up to exceed 3600.

See the WHERE clause syntax chart for the conditions, which applies to both HAVING and WHERE clauses.

The condition must be a Boolean predicate expression. The query returns only rows for which the predicate evaluates to TRUE.

Example 1. Relational query that displays departments with more than ten employees.

SELECT STREAM "deptno", COUNT(*)

FROM "Emp"

GROUP BY "deptno"

HAVING COUNT(*) > 10

 

Example 2. Streaming query that displays products for which there are more than $1000 of orders in the past hour.

SELECT STREAM "prodId"

FROM "Orders"

GROUP BY FLOOR("Orders".ROWTIME TO HOUR), "prodId"

HAVING SUM("quantity" * "price") > 1000;