With trace lines of my two servers I can easily tell that each server is using a little over 50% of their respective memory footprint. When looking at it from a stacked graph perspective, first-off: % usage on the left-hand side is 0%, 100%, 200% which doesn’t make much sense to begin with, and so it’s harder to estimate what the values actually are for each server at-a-glance without mousing-over the graph. It also gives an impression both servers saw an up-tick in memory usage after a particular time.
Those are relatively calm examples, a better one might be this CPU example. Granted, the actual CPU usage below is minimal but it’s still much easier to tell that one server has consistent CPU when looking at trace lines, and most of the peaks/valleys you see in the stacked graph are due to solely to another server.