The other thing I wanted to point out was, it's a high volume transactional environment(Financial data). It's not a warehouse where you load data at fixed time intervals. We can have a user come in and say, I want to process my financials based on my investments, financial transactions entered, say for a month's worth of daily data. At the same time, we have another user who comes in and says, reprocess all my data for the past month(Different set of data, same tables). We would then go and wipe out his financial results for last month(delete) and go back and reprocess data based on some modifications to some transactions he entered effective for last month. We have store results on a daily basis, for months and yearss on end. Like this, we can have about 20 users on the system doing a myriad of processing, and it begins to give you an idea of the volume. Each period(a day) can have results worth hundreds / thousands of rows of data, etc. So, in other words, on demand, hundreds if not thousands of rows of data being deleted, inserted , updated every few minutes...