With SpeedFusion I know its not an additive solution (ie: You have a 200 mbps + 200 mbps connection you don’t get 400 mbps - you get a robust 200 mbps connection) but what I am finding when I have two 5G like connections (very similar ping times and speed tests) SF only uses one of the 5G SIMs to the fullest and only a few mbps from the second connection dragging DOWN the overall speed.
When I run SF only with one active 5G connection (SIM 1 or SIM 2) I am getting the fullest potential of the connection.
It seems overall SF is inadvertently dragging down its overall speed potential.
Turn OFF Dynamic Weighted Balance. When that is on the algorithm will throttle back traffic through a WAN when it’s latency increases.
Non-DWB will allow SF to saturate both links.
In my settings for my SF profile I didn’t see that option.
Its only configurable via the support.cgi webpage.
Speedfusion Cloud Traffic Distribution link on that page will invoke the SFC UI in a way where you can edit that setting.
Ah… I am not using the SF Cloud. I am using a FH Solo.
When combining two DSL connections I was able to use speedfusion to bond two 15 / 1 DSL connections and used speedfusion to get a 25 / 1.5 connection which worked very, very well for me. (I needed the fastest speed possible to transfer a single file up or down, which is why I need speedfusion instead of simply load balancing.)
With cellular, I haven’t yet been very successful at increasing download bandwidth using speedfusion. I’m not sure why, but for me combining cellular download bandwidth has been more difficult than cellular upload bandwidth. Here is the best I’ve been able to achieve bonding cellular:
1 Like
I’m not even interested in increasing the cellular speed. I don’t want to penalize it. I also understand that with cellular there is a LOT in play (latency, ping times, etc) but if I get two connections that is virtually the same the speed just takes a nosedive.
For “regular” speedfusion look at your profile tab screen within inControl.
To add a bit more my SF profile is locally created and not via IC2…
Ok… then first go to your router’s support.cgi web page. Then click the PepVPN Traffic Distribution mode link. That will pop a new tab…where you click your profile and now the option to change this mode will be visible.
I’m not even interested in increasing the cellular speed. I don’t want to penalize it. I also understand that with cellular there is a LOT in play (latency, ping times, etc) but if I get two connections that is virtually the same the speed just takes a nosedive.
For what it’s worth, with bonding (instead of dynamic weighted bonding) I found at my location that bonding two cellular wans is slower than one cellular alone for download speed, so I put the second cellular wan as priority 2 so it will fail over to it if wan 1 goes down but not drag wan 1 down. It has been fairly seamless for me.
Still trying to achieve a little better with cellular bonding as I loved the way speedfusion worked to bond two dsl lines previously.
I would actually advocate for testing DWB here rather than turning it off, it is not the default algorithm and you’d need to enable it on both ends via the support.cgi page.
We have found DWB to work very well where variable links are used and there are disparate latencies or chances of loss / high jitter - but it is not as clear as “this way is always better” but on the whole I can tell you which one is on our default config for anything with multiple cellular backhauls in use at the moment.
Slowing down to speed up is a thing, saturating a link to the point it induces loss and erratic latency does not help the overall performance in a lot of cases.
The detailed PepVPN graphs provide a lot of useful feedback here, it may be worth running some tests over a number of minutes and sharing the graphs.
The WAN analysis tool can also be used to test the end to end capacity between each WAN and the FusionHub outside of the PepVPN tunnel to give you an idea of what that link could deliver.
3 Likes
