Latency variation with SpeedFusion Cloud

Hello,
I am testing my SpeedFusion Cloud setup. It is configured with bonding my LTE connection with WiFi as WAN (2.4Ghz + 5GHz) into one tunnel, and with FTC = Low.

Today I was running a video call from my laptop over that connection and was experiencing slowness of the uplink connection. Please see the graphs below. There were periods of time when latency on all three WANs was going up to over 100+ ms and the connection was dropping packets. As you can see those are strongly correlated. It was fluctuating up and down, as seen on the screenshot, with very distinct changes in the latency. The reason I am posting is that all WANs experienced latency changes at the same time and with the same effect. It makes me think that this might be something on the SFC side of the tunnel or something with the MAX Transit. If it was a WAN-specific, I’d see one of them worse than others.

Any thoughts on this?

Regards,
Maciek

1 Like

Look like you are having a problem your video call. Are you using Zoom, Google Meet, Skype, or the rest of the software?

May I know which SpeedFusion Cloud location you are selected?

Thank you for the quick response.

For this call I used Google Meet. My SFC location is Seattle, and this is where I am physically located as well.

We can take a look of the packet loss. By the way, I am interested to know how you verify the slowness is caused by the SpeedFusion Cloud with the connection below?

Your laptop —wired/wireless—> Transit —SpeedFusion tunnel—> SpeedFusion Coud —> Internet —> Google Meet <— Internet <— Laptop (your friend/colleague)

Your laptop is connected MAX Transit wirelessly?

It is all a hypothesis, and I am looking to prove or disprove it.

Here is what I know/observed:

  • the video call was with a large group. I was able to see and hear everyone, but when I spoke, my voice was garbled. I did not run my video. Multiple people complained that they cannot hear me.
  • During the call I observed the SFC diagnostics graphs that I pasted a screen shot of in the first post. I assume those graphs show the connection between Transit and the Internet beyond the SFC and exclude my local LAN connection.
  • My laptop is connected over wireless from the AP One Rugged, that is wired to Transit. Transit’s AP is turned off. The AP One is sitting 1 ft from my laptop.
  • I watched Transit’s CPU and it was below 20% during the time of high latency.
  • There are three WANs within the tunnel, and they all showed the same latency changes, with that step change, all at the same time and they all showed similar level of packet loss. This is to me an indicator that it is unlikely an issue with the WANs themselves, as it would be strange for independent marina WiFi and a Cricket Wireless LTE to experience the same problem at the same time.

Please let me know if there is anything you can look at from your side.

Regards,
Maciek

Hi @maciek
It looks to me like the WANs where saturated by another device using the internet.
Would you have any statistics on the individual WANs so we can see which one had the packet loss and high latency?
For the Marina wifi, I could understand that been saturated by someone else downloading a large file but that should of not affected this as you had other independent WAN

That’s my idea of using the bonded tunnel that will allow one WAN to take over if the others are too busy. On my side, there was nothing else using the Transit. I watched the stats of the individual connections/WANs (did not take a screenshot though) and all of them were showing a similar level of packet loss all at the same time.

@maciek, look like extra fining tunning is needed for your case. You may open ticket for us to take a look. By the way, can I confirm the Transit is stationary?

Created ticket #20110147

The router is on a boat, but during this occurrence, I have been in a marina and stationary.

Regards,
Maciek