Here is the situation. I have an RB750GL I am using for my gateway router. Customers are under speed plans.. i.e. 1mb, 2mb, etc.. No data limits. Total capacity of my WAN link is 150M/20M. All IP's are static.. 10.10.10.0/24 range.

I have already figured out how to use simple queues to limit data to the plan size by IP address, but....

and this is the part I am having trouble with

I would rather group my customers in pools of 4 (or 3 or whatever works best), set a total limit on that group to 4x the individual cap (so 4 customers at 1mb each, the pool would be 4mb) and then "balance" the usage, so that nobody goes below what they pay for, but if there are spare mbits in the "pool" they can use those to get things done faster..

The reason I chose 4 (and 3 might even be a better option) is I don't want the speeds to be "too fast". i.e. if I've got a customer paying for 1 mb and he/she ends up with 10mb for large periods of time, then when they are actually getting the speed they are paying for, I can see them bitching about "slow internet".... So a little bit of speed boost is fine, but if they want more sustained, they can upgrade to the next tier.

I've spend HOURS trying to figure out how to do this and I just can't wrap my head around it (honestly I'd rather be up on a tower installing a dish or something), so I am willing to pay for the help (nothing in life is free, right)

Still in the startup stages, so I don't have piles of cash laying around, but if you think you could accomplish this and the rates are not insane, I'm perfectly happy to send some $$ via paypal.

You can us a burst limit and set the about of time it has to run at the burst limit speed.

A burst limit is only triggered when the clients are using the max bandwidth limit so if the client doesn't use what is already set in the simple queues as the max bandwidth it will never go over what is set for it.

You will also have to set the burst speed which can only be use if the bandwidth is available.
Then you have to set the amount of that the client can run over the simple queue speed set for it.
(once burst time has elapsed bandwidth will return to the simple queues speed set for it until the burst is triggered again.)

Hope it helps.