The OLE DB Command in Data Flow works on a Row By Row basis. RBAR at its finest. @=)

If you have something huge to update, you're better off trying to figure out how to accomplish it in an Execute SQL Task. What I do is read my data into a Staging table with the Data Flow, then run an Execute SQL Task and, depending on the situation, either use the Execute SQL Task to upload the data to the final destination or return to a second Data Flow to finish my transformations.