Hi,
Balance 310 5G Router, latest firmware.
I have WAN 1 : optical fiber 1 Gbit/s provider
I try a speedtest.net with my computer on Lan1 → Peplink VPN SpeedFusion (FusionHub AWS), I have this Throughput on FusionHub for download :
Why I have same download and upload in Throughput ?
Thanks for your help
Because traffic passes through the FusionHub.
Downloading at 300Mbps from your 310-5G means 300Mbps of traffic has to go into and then back out of your FusionHub, hence why it is called throughput… this is not a problem, it is entirely expected behaviour to see almost symmetric traffic under most circumstances in this type of configuration.
If your issue is with the performance then I’d suggest you post some more details about your configuration and setup so people could look to help you further.
Hello,
Thank you for your feedback, I understand well how Fusionhub works.
On the other hand, I can’t reach the maximum speed proposed by the balance 310X 5G, i.e. 600 Mbit/s, I manage at best to have 300 Mbit/s via the Peplink tunnel
Here is my architecture:
Thanks for your help.
Check the CPU of the fusionhub instance. It may be getting pegged when traffic is above 300Mb.
Screenshot in the initial post shows 21% CPU load with ~600Mbps going through it (300ish in + out) and this is a t3.medium AWS instance so again highly unlikely to be the problem, however perhaps the OP could share the same info for the 310 side as that is far more likely to get saturated 
Let’s be clear about one thing before we go further, that is a marketing figure - so best case throughput under optimal conditions. That said I’ve certainly seen 500+ Mbps of PepVPN throughput out of our 310 units with a single WAN and PepVPN tunnel so I would agree you should probably be getting more than ~300.
Have you used the built in PepVPN speed test? If you log into your 310 and go to the Status > PepVPN page and next to the PepVPN profile click the “>” button you can run some bandwidth tests directly from the 310 to the FusionHub.
Whilst running these tests you can also click the graph button to get some detailed stats, I’d suggest running them for a few minutes and grab the screenshots as they can show if there is any loss/latency on that connection that could be causing the throughput to drop a bit.
I would also suggest using the WAN analysis tools to test bandwidth between the 310 and FusionHub outside of the PepVPN tunnel, this can help you determine that the path between the two devices is capable of the throughput. I would run the FusionHub as a server and the 310 as a client and test TCP and UDP over long durations in both directions.
Lastly I’d maybe bring up a FusionHub somewhere for testing like Vultr or DO, cost is negligible and could help just rule out if there is something else at play here between your ISP and AWSs network (somewhat unlikely, but stranger things have happened).
Hello,
Indeed the CPU load of FusionHub is on my first screenshot.
Here is the balance side:
I agree on the marketing figure, but at around 500 Mbit/s I wouldn’t have wondered.
Test done in download on the server (upload for the balance)
Sur le 310x, le WAN Analyse :
I would do the test from home, it is another provider
Thx for your help
Hello,
I did the test by taking the Balance 310X 5G from another fiber provider.
I have well over 500 Mbit/s down and up in the tunnel!
This means that my current operator provides 1Gbit/s on ports 80 and 443 but on the other ports, the speeds are limited to 300 Mbit/s down and 200 Mbit/s up. Noticing this, I did other tests without Peplink on SFTP, etc… on other ports… and I am limited 
I will try to get some explanations. Do you know what name of filter or rule could be called in the providers the fact to restrict in bandwidth some port?
Thank you for your help on the subject.
@JC ,
Not sure whether QoS/Traffics shaping are the right term to be use by your providers to limiting the different port speeds. Normally this is the feature that we will use limiting bandwidth.
If they are doing it based on simple port/protocol then changing the port the PepVPN is using may help you avoid this.
If they are using more sophisticated techniques and doing actual deep packet inspection (DPI) then you will find it somewhat harder to circumvent.
There is one more possibility - this is very much an unusual example but in the UK there was a problem with one of the big ISPs CPE that used a specific SOC / CPU to offload traffic and it was very poor at offloading certain types of traffic - specifically GRE in this example - if there is some ISP supplied CPE involved here do you have any way to remove the ISP router / CPE to test without it in place, i.e. connect directly from the fibre ONT to the Balance 310?
