The Data Integration Service can push Aggregator, Expression, Filter, Joiner, Sorter, and Union transformation logic to the following relational sources:

Amazon Redshift
Greenplum
Hive
IBM DB2
Microsoft SQL Server
Oracle
SAP HANA
Sybase
Teradata

When you push Aggregator transformation logic to a relational source, pass-through ports are valid if they are group-by ports. The transformation language includes aggregate functions that you can use in an Aggregator transformation.

The following table displays the aggregate functions that are valid in an IBM DB2 relational source:

Aggregate Functions	DB2-LUW	DB2i	DB2z/os
AVG	Yes	Yes	Yes
COUNT	Yes	Yes	Yes
FIRST	No	No	No
LAST	No	No	No
MAX	Yes	Yes	Yes
MEDIAN	No	No	No
MIN	Yes	Yes	Yes
PERCENTILE	No	No	No
STDDEV	Yes	Yes	Yes
SUM	Yes	Yes	Yes
VARIANCE	Yes	Yes	Yes

The following table displays the aggregate functions that are valid in Amazon Redshift, Greenplum, Hive, MSSQL, Oracle, Sybase, and Teradata relational sources:

Aggregate Functions	Amazon Redshift	Greenplum	Hive	MSSQL	Oracle	Sybase	Teradata
AVG	Yes	Yes	Yes	Yes	Yes	Yes	Yes
COUNT	Yes	Yes	Yes	Yes	Yes	Yes	Yes
FIRST	No	No	No	No	No	No	No
LAST	No	No	No	No	No	No	No
MAX	Yes	Yes	Yes	Yes	Yes	Yes	Yes
MEDIAN	No	No	No	No	Yes	No	No
MIN	Yes	Yes	Yes	Yes	Yes	Yes	Yes
PERCENTILE	No	No	No	No	No	No	No
STDDEV	Yes	Yes	Yes	Yes	Yes	No	Yes
SUM	Yes	Yes	Yes	Yes	Yes	Yes	Yes
VARIANCE	Yes	Yes	Yes	Yes	Yes	No	Yes

The following table displays the aggregate functions that are valid in MSSQL, Oracle, and Sybase relational sources:

Aggregate Functions	MSSQL	Oracle	Sybase
AVG	Yes	Yes	Yes
COUNT	Yes	Yes	Yes
FIRST	No	No	No
LAST	No	No	No
MAX	Yes	Yes	Yes
MEDIAN	No	Yes	No
MIN	Yes	Yes	Yes
PERCENTILE	No	No	No
STDDEV	Yes	Yes	No
SUM	Yes	Yes	Yes
VARIANCE	Yes	Yes	No

A relational source has a default configuration for treating null values. By default, some databases treat null values lower than any other value and some databases treat null values higher than any other value. You can push the Sorter transformation logic to the relational source and get accurate results when the source has the default null ordering.

If you configure a Sorter transformation for distinct output rows, you must enable case sensitive sorting to push transformation logic to source for DB2, Sybase, and Oracle.

The Data Integration Service cannot push any function that contains the Decimal data type to a Hive source.