I have an rf link that is not very good, there are plans to improve the link in the spring.
My question is if I set up a tunnel between the two nodes will I have to disable the rf link or will the software automatically use the fastest connection.
Thanks
The software will use the path most likely to succeed in delivery.
This means it will depend upon how reliable the Internet connection is (most likely reliable) and it will choose that path. This may or may not be the "fastest" path.
Should either link go down (RF or tunnel) the software will compensate and use the alternative path.
Any thoughts would be appreciated thanks.
Oh ya I am running the bata4 firmware on 2 bullets.
I would tend to think that the speed is being limited by the CPU. I use a VPN at home for privacy, when i set it up a first i used a RPI b+ and found that it was the bottle neck. I was only seeing 5mbs up/ down and that just wasnt going to cut it for my whole home network to pass traffic thou. After doing some research and choosing to go with PFsence, They Its suggested that one go with a modern intel or AMD cpu because of the CPU intensives nature of VPN Tunnels. With a intel core2 duo and 4gb ram 2 intell pro GB cards, im seeing Full thruput. ~100mbs down and 7mbs up. Comcastic Hi HI
I would conjecture that the node's CPU may be a little under powered to see full throuput on a vpn tunnel.
Has any one benchmarked there tunnel?
I've tested (but not benchmarked) multiple concurrent vtun tunnels running on a NanoStation M2 node with no "end-user" noticeable decreases in speed.
At one time I had 4 clients connecting and 2 server connections. (6 tunnels)
Also, I just check on another node that has a vtun client to a server (1 tunnel) and "top" is reporting the vtun process using 1% CPU.
Overall, vtun is fairly light (as opposed to more feature rich solutions like PPTP, IPSEC, and OpenVPN)