SpeedFusion with WiFi WAN (Starlink) and WISP


Newbie to SpeedFusion here. I am in the Starlink beta program and have been using a WISP connection as my primary work internet. I am experiencing issues on Starlink with VPN and VOIP connections dropping. I would like to take advantage of SpeedFusion to improve that experience if possible.

Since my WISP connection is ethernet only, and the Surf SOHO is limited to a single ethernet WAN, I will need to utilize WiFi WAN to connect to my Starlink router.

Big question - can I configure SpeedFusion to bond the connections, but only route traffic to my WISP connection if Starlink is down? My WISP service is bandwidth limited, so I don’t want to use it unless I need to.

Thank you in advance.

1 Like

Yes you can use SpeedFusion hot failover to send traffic via the WISP connection when Starlink is down. This should be fast enough to keep your VoIP session up and client VPN connected.

From what I’ve seen on my Starlink I get 25+ mini outages a day at the moment and they can individually be anywhere from 15s to 2 mins in length. In filling with WISP or 4G is the only way to make is reliable enough for VoIP.

1 Like

Exactly. I see fairly frequent 10-15s outages which just obliterate my VPN connectivity.

Can you share any tips on this configuration?

Including my current setup based on a guess from suggestions elsewhere. Is this what I need to route everything over Starlink unless it drops, then my WISP?

Screen Shot 2021-03-09 at 15.29.28

I assume YellowKnife is your WISP?

If so:

  • On the dashboard put your Yellowknife connection into Priority1 with the Wi-Fi WAN to Starlink. There will be some data used on the YellowKnife connection as the device tests it to make sure it is healthy and there will be a tunnel built in standby for hot failover.
  • On the Tunnel options disable WAN smoothing.

Yellowknife is the WISP, correct.

I will try disabling WAN smoothing but that seems counter-intuitive to the purpose. My VPN is disconnecting with drops like this:

Perhaps, but WAN smoothing duplicates all traffic over all WANs. You said:

Starlink currently suffers a complete loss of service multiple times a day. When that happens you want to immediately fail over to the WISP connection, but if you are hammering away at the Starlink connection with 100Mbps downloads you don’t necessarily want to be flooding your WISP link with that same traffic too right? That’s why I’m recommending you disable WAN smoothing.

If that isn’t an issue, leave WAN smoothing enabled, but move your WISP connection to Priority 1 on the dashboard otherwise it won’t get used for WAN smoothing.

OK, I see where you are coming from. I work from home and I am the only one one using the Peplink router. Other devices in the home are connected to the Starlink router, but that is separate from how I want to use SpeedFusion.

I still see my VPN disconnecting with these Starlink drops. I wonder if I need to tweak the Health Check Method on the WiFi WAN connection? Seems like it may not be doing a good enough job of detecting a failure.

Hmm. You have a SOHO that normally doesn’t support Hot failover, but I thought that when it was connected to SFC it did. Might be wrong. Suggest you roll back to your WAN smoothing config again, but make sure both your WANs are in Priority 1 on the main dashboard.

I know that you are using your SOHO for 100% of the traffic so you can use SF 100% of the time.

I am using a Balance 20X.

I have both WAN ports up priority #1. Health checks set to 3 failures with 4 second TTL every 5 seconds. Recovery is 1 good check.
(anything shorter will disconnect the starlink during heavy downloads)

I use the default Outbound policy with Priority to WAN Starlink then WAN LTE.

This allows normal traffic to use Starlink and to failover to LTE when the HTTP health check fails.

For zoom, VPN and other high availablity requirements I use an outbound policy for those domains. with Priority: SFC → Starlink → LTE.
with the SF set for both WAN links pri#1. Wan smoothing Normal and FEC Low.

It isn’t perfect, I get a bit of garbled audio in zoom when starlink goes offline. I will have to try some forced outages with various settings. VMware Horizon has no issues.


Thank you for that info!

Can you be more specific about the outbound rule for Zoom? I am trying to determine how I can target Microsoft Teams in a rule.

1 Like

zoom is domain based zoom.us. works for webex as well webex.com.

I haven’t looked at teams yet since work has borked up our teams integration and will probably require that I use the work VM client rather than a native client. (Just tested, yes internal network only)

Tcpdump/Wireshark will usually give you the main DNS domains quickly.

It also depends on if you are using native Teams with servers on prem, or O365 integration.

Check on the " Skype for Business Online and Microsoft Teams" section. You can probably get by with using the domain names + your business domain.

I would assume that if it doesn’t say “ER Yes” that it isn’t really required. Express Route is the method that large businesses will peer internally to microsoft to provide the service. and If it doesn’t go via Express route it is blocked.

If you can’t get a set of rules to work, then I have often just set the IP address of the one computer I need for the SF tunnel in a rule. Which works for home networks.

1 Like

Thank you so much! I will dig around and see what I can find.

I made some tweaks to my configuration based on your feedback, including an outbound rule which routes all traffic from my work laptop over SFC. Since I use Teams (O365) from another computer, I’d like to try to route that traffic if possible as well.


UPDATE: I’ve already noticed a big difference in connection stability. Wanted to thank everyone for the suggestions!


UPDATE 2: My Windows VPN remained connected for over 24 hours. I also conducted a test and unplugged Starlink while I was on a Teams call. No drops!


I have Starlink as well and was thinking about trying to find a used 20X for the same thing, do you have it configured to show stats from the dish? If I only want the hot failover option for LTE ( no bonding) do I need to subscribe to PrimeCare?

Thanks, totally new to all this and its a bit overwhelming

20X’s have only been sold for the past year, so there won’t be a thriving secondary market for them.

If you don’t have PrimeCare and the SpeedFusion ability then your failover will be of the quality listed in the earliest posts in this thread. 10-15 seconds of lost traffic until the health checks kick in and switch lines. Ok for streaming, not for zoom, VPN and other interactive work. I had my SF tunnels not enable yesterday and I wondered why my work remote access kept locking up.

Yes, you can reach the statistics page, but you have to also put in a switch and use one of the workarounds discussed this this thread:

If you aren’t going to use the SDWan capabilities, you would probably be better off with a generic PFsense or OPNsense system. Those support “up/down” on multiple WAN links and static routes on the WAN for the statistics.

1 Like

Hi Paul, thanks for responding, I ended up purchasing a new Balance20X last night, as well as a GS105 switch for the work around - so I will have PrimeCare

That will be a solid investment.
I’m using the USB ethernet dongle WAN for my access to the statistics currently… you can use any of the 3-4 methods discussed in that thread depending on what hardware you have sitting around.