WAN Smoothing - granular usage


#1

It would be very useful if WAN smoothing could be implemented but only specifically used on certain types of traffic.

As an example, identify voice traffic based on DSCP value or precedence marking, source/destination etc and then apply WAN smoothing to that identified traffic, so similar to using QoS rules but for WAN smoothing.

This way smoothing is only applied to the loss sensitive traffic instead of a global setting and duplicating usage for all types of traffic including non critical/sensitive traffic.


#2

Hi Chris,

You can apply WAN smoothing to certain types of traffic by creating multiple PepVPN profiles between two sites (one with WAN smoothing enabled and one without).
After creating several profiles you can create outbound policies for certain traffic to use the PepVPN profile with WAN smoothing (based on protocol used, or source ip address for example).

See: Outbound policies within a PepVPN or Speed-fusion tunnel


#3

Hi Erik, that’s useful of course but another bonded tunnel does take away a significant chunk of usable bandwidth away from the applications (especially on cellular), hence the preference to keep it on one tunnel but be able to control in a more granular fashion.


#4

I wonder what the overhead is on sub tunnels? Most of the VPN overhead is traditionally in the encryption headers I thought. Since each sub tunnel is not individually encrypted (as they transit the encrypted primary tunnel) I suspect the additional overhead of sub tunnels is limited…


#5

Our SpeedFusion engineers have informed me that he only added overhead when using multiple (sub-)tunnels between the same 2 sites are the health check and handshake messages for each (sub-)tunnel.