If a wan link speed changes over time, does speedfusion adapt over time, or does it rely only on the upload and download speeds typed in under wan properties to determine how to divide traffic?
The hard coded speeds set in WAN settings help the Peplink device load balance using capacity based algorithms (ie least used, overflow etc).
Speedfusion uses WAN links based on latency and packet loss measurements. How it uses links can be fine tuned (ie latency cut off and suspension time after packet loss) if needed.
I wonder what the timeframe is for the speedfusion tunnel to adapt to WAN speed change?
Under Network WAN 1 / WAN 2 I started with both set with upload bandwidth: 1100 kbps. DSL 1 and DSL 2 both speedtest .95 to 1.1 mbps individually upload. Speedfusion tunnel speed result is 1.5 mbps upload speed.
As an experiment, I unplugged WAN 2 and plugged in BR1 LTE connection with upload speed of 9 Mbps. Speedtest over speedfusion tunnel still clocked 1.5 to 1.6 mbps after 5 minutes. (many speedtests done in those 5 minutes to move data over speedfusion tunnel.) It did inch up from 1.5 to 1.6 mbps after 5 minutes. I went into WAN settings and changed upload bandwidth to 10 mbps for WAN 2 and then speedtest over speedfusion tunnel measured 4.5 mbps upload speed over the combination.
Tested some more, and it doesn’t seem to have any effect on the speedfusion tunnel speedtest so long as the upload bandwidth: number entered is greater than the actual wan speed, but when the number entered there is less it seems to impose a limit.
The Upload / Download bandwidth you set in the WAN settings page is always the maximum limit of that WAN, so in your case, if you set WAN2 to 1100kbps, and plug a BR1 LTE with 9Mbps capability to WAN2 via Ethernet, the maximum speed will be limited to 1100kbps.
Thanks both of you for all the info. (and sorry for so many questions as I get started and explore a couple different directions to head for the best solution for a difficult location.)
Follow up question: with speedfusion set to the default bonding (traffic distribution: bonding) of WAN 1 + WAN 2 to a fusionhub, is there any disadvantage to setting WAN 2 upload/download limits on the WAN settings page to the maximum that WAN might get (even on a few-day span where actual performance of that WAN is much less than that maximum?)