For each customer-product combination there can be multiple prices to be calculated: the actual price plus zero or more future prices.
These data will be imported into BI to make comparisons for past, present and future turnover per customer (group) and/or product (group)
This results in approximately 500.000 prices to be calculated (only current price, no future price). Approximately 10 prices per second (in a single tread) can be calculated which would take up to 14 hours for the calculation. Therefore, parallel executing is needed and caching wherever possible.
On one AOS approximately 10 parallel tasks can be executed, this results in approximately 70 prices/sec per AOS which results in 2 hours of calculation (tested on virtualized SQL server and AOS).
If we use 4 AOS servers then we could go to (2×60 + 2×40) = 200 prices/sec. which would reduce the calculation time to less than one hour.