I have an application where I need to tunnel 1500 byte packets through an interface with a 1500 byte MTU. I’m hoping to use a CCR1009 at each end of the tunnel. The tunnel will be EoIP or some other GRE-based IP tunnel. Any packets over 1476 bytes will need to be fragmented by one of the routers and reassembled by the other.
What kind of performance (in packets per second) can I expect from this situation? Is the packet reassembly process (the most CPU intensive part of this operation) multithreaded? Any suggestions for improving performance here?