mpls over microwave

Always used Ethernet handoffs on the radios to keep
things simple.

One more add:

Properly engineered, fixed wireless links can have better-than-wireline
availability. Two jobs ago, we had customer links with zero dropped
packets in 5 years, which is outstanding compared to most copper-based
services.

Properly engineered, however, is the key. Make sure whom-ever is building
your links looks at vendor specs, builds a real link budget (including
losses from connectors, cable, grounding, etc) properly weather seals
everything, and try to get at least a a 20db fade margin if you can. If
the things I just mentioned are confusing to your RF guy, you might want
to get outside help.

Make sure they can know the models for propagation as well. At lower bands
fading caused by K factor change can be a bitch and is the predominant mode of
fading. on the higher bands (>11GHz) rain fade is the predominant mode of fading.

If you want your network to have 99.999% uptime, the links need to be
engineered for 99.9999% uptime.

Make sure the consultant knows Pathloss5 and is not using a vendor's program
for that (orthogon/motorola/cambium/whatever they call them now) loves to push
their own shitty program.

If you're using a dynamic data rate radio (most are), ensure you engineer for
highest data rate you need. If you run the calcs and it shows 99.9999999% at
bfsk but you need 128qam to get your full data rate, what good is it?
congestion do to link down modulation is still an outage.