G
George Potemkin
Guest
To simplify a picture let’s assume that database has only one latch - MTX latch. The one I don’t like because when MTX latch is a bottleneck I don’t have any suggestions for the customers how to improve performance. Let’s assume a box has 128 CPUs. Will we get the stronger issue with Cache Coherency if we will split the whole table space, let’s say, in a hundred separate databases with their own copy of MTX latch?
Continue reading...
Continue reading...