How many products are you talking about? Because if it is less than 10 million, don't even bother optimizing it that way.
What you are talking about is basically a separation scheme. This is used for really large data sets. Less than 10 million do not come close, although each element is massive.
If what you do reads a lot, few write that there is no benefit from this kind of splitting.
Do not try to optimize the problem until you have a problem.
source
share