I selected sorted input but the workflow takes the same amount of time as before.
You cannot use sorted input if any of the following conditions are true:
The aggregate expression contains nested aggregate functions.
The session uses incremental aggregation.
Source data is data driven.
When any of these conditions are true, the Integration Service processes the transformation as if you do not use sorted input.
A session using an Aggregator transformation causes slow performance.
The Integration Service may be paging to disk during the workflow. You can increase session performance by increasing the index and data cache sizes in the transformation properties.
I entered an override cache directory in the Aggregator transformation, but the Integration Service saves the session incremental aggregation files somewhere else.
You can override the transformation cache directory on a session level. The Integration Service notes the cache directory in the session log. You can also check the session properties for an override cache directory.