Data Mover jobs that use Teradata ARC have the unique ability to use the Multi-ARC functionality of Teradata ARC. With Multi-ARC, multiple Teradata ARC processes work together on the same workload. The pairing of a Teradata ARC archive process with a Teradata ARC copy process is referred to as a stream. The number of streams can be specified using the data_streams parameter when creating the job with the command-line interface create or move command. If the user does not provide a value for data streams, Data Mover will dynamically determine the number of streams to use based on multiple factors, including number of AMPs, size of objects being moved, and so forth. This dynamic value is a good starting point, but advanced users may find more ideal settings to use for each situation.
Multiple streams are limited to running on a single agent; therefore, using multiple streams does not improve Data Mover job performance if the bottleneck is the workload of the agent or network bandwidth. Using too many streams can lead to performance degradation; however, if the agent has resources to spare, increasing the number of streams can maximize the performance by using the extra bandwidth and processing power of the agent. Only increase the number of streams when copying large tables. Use only one stream if there are small tables in the Data Mover job (even if there are many small tables).