BGP Mesh Configuration / Full Routing X Devices with low cap

RouterOS Forwarding Protocols
1

Hello Friends

I have a problem to solve, and I need a ligth in the end of tunnel.

I have 3 BGP servers. These 3 BGP servers make a full routing with another 3 different peers (3 servers X 3 links = 9 different links). I will call these routers the Router A, Router B, and Router C.

The problem is that in this servers I have a full routing with these operators, generating a lot of routes.

In my case, I have RB 433 and RB 600 acting as AP and Router in a remote location.
The problem is: if I set a default route to router A (example), but the best route is in router B, I have a re-traffic in my network (the packet arrives in A, is re-transmitted via wireless to B). This scenario to me is not good, because the link is too far and I have only 1/2 MB of throughtput, and any single packet retransmitted generate a desnecessary traffic.

The problem is, if I put a device such as RB 433 ou RB 600 in the BGP mesh, the large number of routes consumes all the resources, and the router reboots.
My question is: is possible set a default route to one of routers (A, B or C), and propagate to device only the best routes in B and C?

I believe that isn`t possible, but, in all the cases, I post the question.

E.G.: What is the “route reflector” or “BGP confederation”?

2

if you really have 9 providers you probably dont need to be exchanging full routing tables with all of your routers. Just accept customer routes from each peer which should be much smaller number. That way your other routers aren’t receiving a million routes from each other. you can then use routing-filters to set properties on the routes learned to make them less or more viable.