I currently have a fully routed network and my 5ghz backbone links are setup as follows.

1. AP's are setup in mode "AP-bridge"
2. Clients are setup in mode "station"
3. SBC's are RB333's RB433's, and RB600's.
3. Cards are all R52/H's, WLM54AG's.

I'm not using anthing else just very simple P2P links and I can get on properly setup links 22mbps UDP when doing throughput tests from client to AP.

As I stated above I am getting 22mbps UDP across each link. The problem is that when i do a bandwidth test THROUGH one of that backhaul sites to another I see throughput cut generally in half. At first I thought it was self interference between cards mounted so closely on the boards. But all boards that have 2 backhaul cards on them are running in different bands. 5.8ghz and 5.2ghz. Surely there is no interference between the two at 600mhz apart.

Now another factor I have to this equation is that one of the sites has a single radio per RB433 (other two slots empty) and this site acutally will sustain the 22mbps UDP going through it because the backhaul cards have their own dedicated RB433 just for them.

Now my question is as follows:
Do I have to install a separate RB433 board for each radio (surely i hope not) in order to get the full throughput 22mbps UDP going though each site? Or is there something I am doing a wrong way? Is there a better way than what I am doing? Is this a problem from having a routed network? Because I used to have all Tranzeo 5ghz stuff on my backhauls and that bridged backhaul design was able to do 22mbps UDP through every site.

Ok i found my problem, one of the backhauls had a bad card. It was a senao 8602, had terrible ccq on every channel. Replaced with a CM9 and now i can get 24mbps UDP through from both ends of my network. I turned NSTREAM on for the first time ever and WOW! 37mbps UDP from one end to the other I am impressed! Mikrotik has a great product here!