What if shield absorbs were capped per second?
Shields cast on allies refresh (and stack) on the target every second for an amount equal to tooltip value divided by shield duration.
210 / 5 = 42 shield per second
110 / 3 = 37 shield per second
170 / 4 = 43 shield per second
220 / 4 = 55 shield per second
125 / 1.5 = 83 shield on application, half that on refresh
215 / 2 = 108 shield per second
The idea here is to make shielding less of a total counter to burst damage and more of an effective health buffer over the shield's duration. Reactive shields would be weaker, proactive shields would be situationally weaker, and shields in extended fights would be power neutral.
As you can see, this also normalizes the margin of error in casting a shield. Sona's shield has an extremely short duration of 1.5 seconds, meaning that an effective cast has a 1.5 second window in which the shield target needs to take damage. Janna's shield has a window of 5 seconds, which is a very long window to play around and a very large margin of error on an effective cast.
Thoughts? I personally play a lot of shield supports, and I would be ok with something like this.