Metal is a very good equipment , but in my opinion you used it in wrong place !!
Metal is 31 dbm high power radio , perfect for long range and long distance but not good for hotspot or multipoint links.
Metal has only 400 mhz CPU and 64 MB ram.
you should use use a board with powerful cpu and ram like RB 435G or RB 800 , with a stable wireless Card , like XR5 , SR71-15
if high power equipment like Metal use in short link its high power makes your link unstable and poor throughput.
most of the time more power = more distance not more throughput !! if you can not replace Metal , reduce its power , but when many users connected to it , its CPU and Ram can not support all of them , that's why it hangs !
For hotspot or any indoor wireless, I'd never use these radios. I would use any radio at 17-20db Tx, and 3-6dB antennas. It's very likely that I wouldn't use any MT solution for the APs in such an environment.
For any PTP or PTMP solution, why would you need more CPU or memory? Look at the RB711G-5HnD. Same CPU, half the ram. It pushes 160Mbit/s with ease. PTP/PTMP makes no difference, the time slotting on the radio happens outside the CPU.
You are right that the SHPn devices (Metal or mPCI) were marketed for long hauls, but they'll also let you reach the limits in a PTMP configuration.
I would never do a 2.4Ghz long haul PTP, but the math is pretty simple. Start with a 29dB antenna and a 2dB cable loss, restrict your TX power to about 23dB, and you end up with a max EIRP of 50dB.
In a PTMP application, start with a 15dB antenna, take out 2dB line loss, restrict Tx power to 23db, and you're at the 36dB max EIRP.
Compare this to the R52Hn. 2db less tx, and 6db lower Rx. This means that the SHPn has a total of 8dB better margin to work with.
I do love the XR5, but this doesn't support 802.11n. The SR71-15 at MCS7, is 19/-75. This is gives a margin that is again 7db smaller than the SHPn, and gives a max EIRP of just 32dB, 4dB short of the max EIRP.
Here's a real life case study... I had a cell driven by a cm9 with a 12db omni. We had nothing but trouble reaching clients above the AP (the antenna has a 7* beamwidth with a 3* downtilt). Anyways, we maximized the potential of this site by replacing it with a R2SHPn and a 9db Omni. This give us an EIRP of 29dB @ MCS7 and great vertical coverage (-7 to +7), and a much better Rx sensitivity. If needed, we could have even better vertical coverage with a lower gain antenna.
Now, tell me again why the SHPn is unsuitable for a 2.4Ghz AP?