Running Speedfusion/Fushionhub on TCP protocol

As i understand from PepLink Team that Speedfusion run only on UDP . It is problematic since most Mobile internet provider has Qos and queuing mechanism on UDP thus experiencing alot of drop packets. As we are aware that UDP is stateless protocol.Thus such loss packets lead to instability and render speedfusion useless in most of countries where it is deployed. TCP proven as much more reliable protocol even on the expense of high bandwidth usage but it will be more resilient to packets drops and ability to recover those packets. I strongly urge your team to add the TCP option to DataPort in Speedfusion/FusionHub in order to achieve reliability. Otherwise , bonding and WAN aggregation is useless in many countries.


We have currently deployments in Panama, RDC, Egypt and other places where reliability of the mobile network is a luxury we don’t have. And SFusion works.

I understand your current point of view, it is a common reaction to the first tests and exchanges with SFusion devices. Give it a bit more testing and fine-tuning and you should see the benefits coming out of it. If you have specific questions or use cases, this forum is full of well-intentioned peplink architects that can guide you.

The fact that SFusion is UDP based is, according to me, its strongest asset. It works whatever you try to push into it. Encapsulation of IPSec tunnel, done. Voip consolidation, done. Interfaces with variable performances, done. Etc…

1 Like

This is acknowledged. We have seen several cases that a few ISPs are using some kind of QoS mechanism and TCP on these networks perform much better than UDP.

One thing to note here is, no matter TCP or UDP, they should perform equally well if there is no hidden QoS from ISP (this is the most common case we have seen), it’s not related to the default protocol behavior that whether it has it’s own flow control or packet loss retransmission algo.

I have put this on the SpeedFusion roadmap now, but I don’t expect it can solve all the QoS problem from ISP, because when they can lower UDP performance in some way, they can also only boost HTTP / HTTPS performance and lower other TCP performance which is not web browsing. Stay tuned.


Hello Mr. Venn,
I am not here to assess sucess stories of Sfusion and other pepwave products. I came here to solve a technical problem that i faced. The system has sever packet loss and i have been trying to solve it for over a month. I have your engineers on top of it and they couldn’t be able to solve it. so it is much more complicated then you think. I am tying to be helpful and i provide tangible facts that things working fine at TCP level mainly TCP 443. What is mentioned by Below PepLink Team ( Mr. Steve ) is the exact and right answer.

Hello Steve,
I agree perfectly with what you have mentioned. TCP is under Qos but it is much more resilient. the perfect performance is under TCP port 443 or other major production ports.Since ISP don’t put under heavy queuing as you have mentioned. Most of Tunneling services are using TCP for that reason.