You can set up a FusionHub at a data centre anywhere in the world very quickly.
1 Like
FusionHub solo is free but your cloud infrastructure where itâs deployed isnât.
I havenât looked at te install requirements but I assume requires a Linux VM of some kind.
So using this to run automated speed tests would also incur bandwidth charges from your cloud provider depending which tier and plan your using.
I guess this is also only useful for connections less than 100mbps.
Anyone running this in the cloud and could give an idea on costs?
Fusionhub Solo at vultr.com $5/month with 1TB of bandwidth allowance. Not sure what speed connections they provide but I have ~400Mbps at home and it lets me pull that:
5 Likes
Hmm⌠that is an incredible deal. Seems an Azure basic Linux VM with 1TB monthly bw is $103 USD unless i didnât input the right values in their pricing calculator.
Just to build on the experience of @MartinLangmaid, hereâs a speedtest from my office (1 Gbps fiber - my ISP) to a vultr fusionhub solo node:
It works 
Z
3 Likes
In the âwhy notâ category of things:
I have deployed a fusionhub solo instance on a vultr.com farm in Silicon Valley and set it to respond to peplink speedtests. Iâll let it run for a bit for all to enjoy.
I have seen it deliver 900+ Mbps download speed.
The address is 144.202.99.195 (speedtest.apeskrekk.com).
Enjoy
Z.
5 Likes
You can also run the WAN Analyser between FusionHubs in different data centres!
I have seen 5Gbps between a FusionHub in London and a FusionHub in Paris. Both FusionHubs where with different cloud providers and small instances with 2Gb of RAM.
1 Like
I have a Vultr instance of Fusionhub as well. What do you mean when you say you âset it to respond to peplink speedtestsâ? Do you mean that you logged into your Fusionhub server and set WAN Analyzer to act as a Server? Or is there something else I should know? Thanks.
I use Vultrâs ability to download large files a lot. I often initiate several downloads in order to see if the servers or Peplink client scales based on traffic. The Peplink Max Transit is ALWAYS the bottleneck since the CPU isnât very powerful. 100Mpbs (not) without encryption, 60Mbps with encryption. There is tons of degradation on the CPU if the router has to maintain the tunnels. I set the Fusionhub to Server and the Peplink router to Client, then run WAN Analyzer to see the bandwidth stats while downloading the large files from here Vultr Looking Glass, here:
Thatâs correct.
Z
1 Like
Iâll ask again. Why doesnât Peplink provide a simple to use speed test that uses two Peplink routers, one source/target? Your customers have been asking for 3 years.
But I would still like to see a standalone built in speed testâŚ
1 Like
Me as well. I want to see how our traffic performs from my client units to our end-point. This helps determine if I have issues at the client side. The above doesnât really make sense to me on how to use it. We have one Public IP all the rest are DHCP connections.
Go to your end point / central location that has the public IP, click the âAs a Serverâ button. Then go to a client device and click âAs a Clientâ and paste the IP of your central unit in against each WAN link you want to test and it will run a point to point speedtest.
1 Like
@Rexymav From where to where? The issue is always that public speedtest services (like speedtest.net / fast.com etc) have end points that might be near or far, busy or idle, ISP traffic optimised or not. Typically what we really want to know is what is the bandwidth from point A to point B. That is the point and purpose of the WAN analysis tool.
1 Like
Thanks. Issue is that it gives me only the one WAN link at my end point. I have one static address I can use (yet there are multiple WANs connected) to my end point. I of course can test from the client across the multiple WANs (all DHCP) at that end.
Exactly what the Peplink Broadband Speed Test executable offers but built-in. Have you ever used that?
1 Like
Of course. But I donât see the value in it.
The concept of a single point in time network speed testing when you donât own both ends of the test - or the network in between, has very little value apart from acting as a loose guide as to what bandwidth might possible some of the time. I just ran three consecutive speed tests on my cable connection and they were different by more than 100MBps. What does it prove?
This is a report from a sam knows bandwidth test box on one of my Starlink deployments: That is the right way to do bandwidth testing. Over time building a comparative dataset that shows any trends.
Now if Peplink let us run a service on a PC or a raspberry pi that perform speed tests and uploaded results to IC2 for comparative review later - then that is something I could get behind
Anything else for a âgeneral internet speedtestâ feels like it would end up being misleading in the same way speedtest.net and the others often are.
2 Likes
The value is a quick and easy method of testing remote devices for people who may not have their own device to use as a server. Just dial in via IC2 and have at it!
I do however agree with everything else you said in particular, uploading results to IC2. What I am saying is why not have this function embedded in the router code instead of requiring an external device to run it as a service?
One other thing is that the Peplink Broadband test allows large tests to average out short term speeds from the test sites you mentioned.
2 Likes