Jobs that copy large amounts of data using Teradata PT API can distribute the copy tasks for a single database object across multiple agents. Performance is improved because Data Mover draws on the combined resources of data streams associated with each agent. When an agent uses multiple data streams to process a task in parallel, the object is copied faster.
The following restrictions apply when using multiple agents in parallel:
- More than one agent must be installed in the Data Mover environment.
- The job can only use Teradata PT API.
- Jobs that use Teradata PT API only copy data in tables.
If max_agents_per_task is not specified or specified with no value, Data Mover dynamically sets the value for optimal performance.
The following parameters file for the create command assumes that the Data Mover environment has five online agents. Settings in this file are valid for jobs that use Teradata PT API.
<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<dmCreate xmlns="http://schemas.teradata.com/dataMover/v2009"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://schemas.teradata.com/unity/datamover.xsd">
<job_name>floyd_dmdev_create</job_name>
<source_tdpid>floyd</source_tdpid>
<source_user>dmguest</source_user>
<source_password>please</source_password>
<target_tdpid>dmdev</target_tdpid>
<target_user>dmguest</target_user>
<target_password>please</target_password>
<data_streams>4</data_streams>
<max_agents_per_task>4</max_agents_per_task>
<source_sessions>4</source_sessions>
<target_sessions>4</target_sessions>
<log_level>0</log_level>
<database selection="unselected">
<name>dmguest</name>
<table selection="included">
<name>orders_2010</name>
</table>
</database>
</dmCreate>