DB2 Version 9.7 for Linux, UNIX, and Windows

In-database analytics with SAS embedded process support added

You can use an in-database approach to analytics in the data warehouse by running the SAS embedded process (SAS EP) on the DB2® database server.

The ability to dynamically score, or to run regression, clustering, neural net, and other analytic algorithms within the database server environment-thereby pushing the processing and deep analytics to where the data lives-reduces the latencies that are associated with moving data, and increases flexibility in end user analytic capabilities.

SAS Scoring Accelerator for DB2 enables the scoring process to be done inside the database and does not require data movement. Prior to Version 4.1, SAS Scoring Accelerator for DB2 translated the models that were developed by SAS Enterprise Miner into scoring functions that could be deployed inside the DB2 data server. The scoring functions could be used in SQL statements like other DB2 user-defined scalar functions, providing the following benefits:

Reduced data movement and storage requirements
Better data governance (most of the data stays inside the database)
Greater computational power from the relational database management system (RDBMS)
Better productivity through a shorter cycle of ideas to product

New enhancements in SAS Scoring Accelerator for DB2 Version 4.1, which eliminate the need to register scalar user-defined functions (UDFs), make the process of deploying and running scoring models in DB2 more dynamic, and result in better performance when executing models against large data sets. Scoring models that were developed for SAS Enterprise Miner can be deployed in a DB2 database in their native form. The models can be referenced and used in SQL statements through an analytic expression.

The ANALYZE_TABLE expression, which you can specify on the table-reference clause of a subselect, enables you to efficiently execute scoring models. For more information, see "table-reference clause" or "Analyze table expressions" in the "subselect" topic.
The DB2_SAS_SETTINGS registry variable enables the SAS EP. Use the db2set command to configure its settings. For more information, see the "Miscellaneous variables" topic.
The SAS EP library is loaded and run in a fenced-mode process named db2sasep. In a partitioned database environment, this process runs on each database partition of the DB2 instance. For more information, see the "db2ida - Stop or restart the SAS embedded process command" topic.
TBFUNC is a new explain operator type.
When SAS queries are processing, DB2 agents might have to wait during communications with the SAS EP. This can occur when sending data to the SAS EP or when receiving data from the SAS EP. Two new in-database analytics wait times in the time spent hierarchy (ida_send_wait_time and ida_recv_wait_time) provide visibility into the impact of these waits on overall system and query performance.