I've been reading a few posts where people with multiple ISPs want to route different policy-based traffic through specific WANs. The official answer from Untangle seems to be this can only be done by source or destination IP. But I think I found a way to approximate it closely.

As an example, say you want all YOUTUBE traffic to go out WAN2.

You can use Event triggers to tag a host (sessionEvent.CClientAddr) that matches application YOUTUBE. Then, set the tag lifetime to 3 seconds. In WAN Balancer, route hosts with the matching tag through WAN2.

All traffic from the host during those 3 seconds will be routed through WAN2, however immediately after the tag expires, traffic resumes through WAN1. But here's the catch ... since the Youtube session for WAN2 is still established, it will continue routing through WAN2 after those 3 seconds, even though all other traffic is going through WAN1.

It's not perfect as all new sessions during those 3 seconds will go through WAN2, but it's worked fairly well for our use case. Curious to hear others' thoughts on this?