I’m doing some optimization to the outbound policies and so on at one of my sites. They are running a B1Core with a SF licence and an extra WAN licence. The other end of the SF tunnel is a FusionHub appliance. Through experimentation, we have discovered that the total downlink bandwidth available by bonding WAN1-2 and WAN3-4 separately through the “Multiple Tunnels” option is greater than bonding all 4 together. This is only observed when we use the built in load test for TCP packets, not UDP. I believe this likely has to do with the latency mismatch. 1-2 are more alike and 3-4 are more alike.
In the past, that remote site generally had all traffic routed via FusionHub over 4-WAN SpeedFusion. Is there a way to leverage our observation about the multiple tunnels without using outbound policy like this post describes? Something like a 1:1 (or 3:2 in my case actually) weighted balance between the two tunnels without having to pick a specific type of traffic?