A lot of the complaints I've seen about castes vs non-casters is from posts or comments that start with "The rules say...", or "I played one session and...".
These are often either unique or rare occurrences that people get up in arms about, or more likely, just theory being tossed around without seeing what happens in actual play. Something might look horrifying on paper... but plans never survive first contact with reality, because abilities and spells can't be looked at in a vacuum.
I have played two major 3.5 campaigns (well more than one in each case but groups). In the first one we were nearly pure core 3.5 and I never saw the casters dominate. Usually it was the Paladin, of all things. Only in the last couple of encounter of the campaign (level 16) was that looking like a less clear statement but it is hard to judge from a couple of encounters and the level 15 Paladin was still doing exceptionally well in the previous level's encounters.
Later on, I played a late 3.5 game with roughly 25 of the books allowed. My experience was that the casters completely ruled the field. Both in a high level game that I started in and in the second part where we began at level one. It frustrated the DM who claimed that rogues and warriros should still be viable but the whole party started to end up as casters. Part of it was that it was a very high danger campaign and the evasion type spells where literally life savers (dimension door, silence and invisibility).
So I have seen both happen in long term play. I am convinced that blending (many prestige classes and odd prestige class that directly compensated for caster weaknesses) were the trick in the scond campaign, though, as people found all of the ways to get huge saves (very important), melee ability and hit points despite being casters.