I understand that an ETL is run at the same time for all the institutions in a specific data center. Institutions in a specific data center should be from a similar geographic area. If that is not possible, the servers should be grouped by geographic regions spanning approximately 3-4 hours time differences.

The time of the ETL run should be optimized across clients in the given geographical area. For example, if the clients are largely from the United States, the ETL could run close to midnight Eastern, which would be close to 9 PM Pacific.