Benighted

Maybe someone in here can help with this.
We have an IVR with the following components:
- 2 x MPS 500 with 30 channels each
- 2 x Speech servers
- 1 x CCT 5.0 (+1 cold standby)
- 2 x xMP servers hosted offsite (contains the VXML call flow logic)

Today when a customer calls us they get the IVR. What we do is that we simply have a ROUTE CALL xxxxx in the beginning of the script, and Symposium then transfers the call to one of the MPS's. This method causes some issues:
- the calls are distributed randomly between the MPS's. We would like the first 30 channels on MPS1 to be filled up first. If the amount of calls exceeds 30 then MPS2 should start taking calls.
- if one of the MPS's are down for some reason the call can be transferred to that MPS anyway. This causes the customer to get a technical error, and get transferred to agent. We want to check if the MPS is up before transferring the customer. If the MPS is down the call should be sent to an agent through standard scripting.
- a call can get transferred to a faulty port on the MPS causing the call to hang in ringback or silence. We want to check if the port is ok before transferring the call. If a port is faulty the call should be sent to an agent through standard scripting.

All of the above is solveable through Symposium scripting, and potentially some changes to the CS1k. This is according to Nortel.

Currently the IVR is set up as one IVR ACD DN in Symposium. It's called MPS_IVR and has a corresponding number. I don't know why it's set up like that because in my head each MPS should have it's own ACD DN? This way we would be in more control over where we send the call.

Do anyone here have any suggestions on how we can solve the challenges mentioned above?