We are looking at ways to improve load performance to an info cube. Can someone advise on whether number range buffering will improve performance if as part of the process chain we are actually dropping the data before we load data? Thanks