Balance is one of the big goblins of game design. Over the past decade or so, its position as a sacred institution has been (thankfully) tarnished, so it's no longer an automatic assumption that every character needs to be balanced with every other character in a strict technical sense. More than any game, I point to Eden's Buffy RPG for driving this point home, with an explicit power level split offset by other play elements. But the funny thing is it's an old idea. One of the cornerstones of old school D&D, the magic user, was based on a foundation of imbalance. Depending on level, you could expect him to be far less or far more effective than other party members.
Now, like all such ideas, there's a bit of a pendulum effect to it. Once you get discard the idea of effectiveness-based balance, it's not a long trip to treating it as a bad or restrictive thing - something to be discarded. I understand that impulse, but it's overkill, and to understand why it's worth pulling back a bit to examine the thinking behind balance.
See, Balance is a means to an end, and that end is this: everyone playing should have a fair chance at having a good time. If you have wildly disparate power levels in a game with a strong combat element (think D&D) then you end up in an Angel Summoner & BMX Bandit(vid link) situation, where one character solves problems and others get to watch. That's a bad outcome based on the fact that it's a less fun outcome.
It's with this in mind that a lot of models have been create to support balance. As another example, if a game has other avenues of play than combat (like social or political), the idea became that you could achieve balance by allowing characters to excel within their specific arena. The combat guy gets to shine in fights, the talker gets to shine in social situations and so on. This can work, but it takes a LOT of effort. One arena (often combat) can overwhelm the others if the game's mechanics lean that way or if the stakes are higher. A good GM can juggle this, but doing so is almost always a function of GM skill, and that's not a great thing to depend on in a design.
An interesting corner got turned when some games opened up a different venue and moved the issue of balance onto the player. The idea, generally speaking, is that every player has equal (or at least equitable) power or authority, even if their characters do not. This model can range from Buffy (Slayer is more powerful, supporting characters get more 'drama points') to full on hippie ideas like giving players narrative authority.
None of these solutions work in every game, but I think the last one is very informative, even if its never used. The emphasis that it's the players who need equal time is of critical importance because it comes back to the original problem: keeping everyone engaged. It's easy to get bogged down in the details of a specific power or specific rule and forget that the reason you're doing all this is to keep your players engaged.
Now, why is all this necessary? Can't a good GM fairly distribute spotlight time at the table? Well...not really.
It's not that the skill doesn't exist, but to do it well we need to be much better at self assessment than any of us can reasonably expect to be. As a GM, we're going to be drawn to the problem cases or the things that we think take things in an interesting direction. Those are good impulses, but they mean we are vulnerable to spotlight hogs, and we're going to misjudge how fairly we distribute the time.
All of which is to say that you want to have some manner of focus balancing mechanic, even if it's a simple as "This is Bob's spotlight episode." Mechanical balance or distinct roles are perfectly valid ways to handle this (something 4e thrives on), it's just not the only way. So take a look at some of your other games and think about how they hand (implicitly or explicitly) keeping everyone at the table engaged. You might pick up a trick or two.
1 - Ars Magica also had a profound disparity, but its handling was still overall equitable.
2 - This is, in my mind, why 4e is designed to be a pretty weak system outside of combat. The balance is _explicitly_ within the scope of combat, and stepping too far outside that sphere risks disrupting the finely tuned machine.
3 - If Primetime Adventures 3e were out right now, I'd plug it here. But it's not, so I can't. Instead, I'll say this: if you ever get a chance to read a copy of PTA (any edition), stop and look at the spotlight rules. They're genuinely brilliant.