Parallel Execution Plan
As shown in Figure 1 on page 20, the Teradata PT processor is the nucleus of the Teradata PT infrastructure. All user requests for importing, exporting, or updating data are submitted to the Teradata PT processor as a job script using high-level, nonprocedural Teradata PT statements. (See Teradata Parallel Transporter Reference for a description of the Teradata PT statements.)
The Teradata PT processor parses the statements in the job script, validates their syntax, and creates instances of the specified operators, connected by data streams and a plan for coordinating their execution, called a parallel execution plan. During job execution, the Teradata PT processor:
1 Uses the Teradata PT metadata to establish the data streams between operators
2 Validates the user’s access privileges for the data source and destination objects
3 Establishes the optimum path for accessing data within the enterprise
As shown in Figure 2 on page 25, the Teradata PT processor calls the five operator entry points of each operator defined in the job script in a coordinated manner that optimizes job data flow. These entry points (Initiate, Execute, Checkpoint, Restart, and Terminate) are implemented as functions of the Teradata PT operator interface.
Note also that the Teradata PT processor can launch multiple instances of any operator defined for parallel execution to scale up performance through parallel processing.Figure 2: Parallel Plan Implementation