Estimating the Number of Threads

Begin your estimate of the number of data preparation threads by multiplying the number of CPUs/cores on the database server by 2.5. This value should achieve the best possible performance for data preparation tasks as long as the database server is available for exclusive usage by Related Extract for the total extract duration, the DBMS the extract is running against is not sharing access with other applications, the extract specification does not include tables that are extremely large, and the database server has available memory of approximately 1.5GB-2GB RAM per CPU. If any of these conditions are not met, reduce the number of data preparation threads accordingly. Further, if the Running Total value in the File-AID Processing Status window doesn't vary for a period of time, or all CPUs on the database server are close to 100% utilization, or when OS- or database-specific performance counters show significant congestion, reduce the number of data preparation threads accordingly.

Begin your estimate of the number of data processing threads by multiplying the number of CPUs/cores on the execution server by 2.5. This value should achieve the best possible performance for data processing tasks as long as the execution server is available for exclusive usage by Related Extract for the total extract duration, the execution server is not sharing access with other applications, I/O is scalable for the disk drive used for writing the extracted data, and there is limited or no impact of Data Privacy processing. If any of these conditions are not met, reduce the number of data processing threads accordingly. Further, if all CPUs on the execution server are close to 100% utilization or when disk I/O counters show significant congestion, reduce the number of data processing threads accordingly.