Oof, that’s quite low isn’t it. It may look low compared to the speedtest but it is also a very repeatable test - constant server, identical test file each time etc. I know those Vulr test servers can shovel data at ≥1000Mbps quite easily and that is even to things outside their network so its unlikely to be the source in that respect.
I’m assuming that without the SF enaled the Vultr download goes a lot faster?
I do notice in your graphs above when you ran the Speedtest most of that traffic was on your Verizon link, but the HTTP transfer is pretty evenly spread across the Verizon and AT&T.
A couple more quick tests:
Fire off a few copies of the download at the same time from Vultr, does the aggregate speed go up, mabe leave a minute or so between each one so we can see the if having multiple obvious connections gets you better results here.
Try this with just Verizon or just AT&T enabled too, but leave the SF enabled, just drag one WAN into standby on your TST dashboard is probably the quickist way to do this.
On the PepVPN status where you can run the bandwidth tests there is a button that will open a new browser window with a lot of extra graphs, could you set that to “All wan to wan” and leave it running whilst testing and then export it and upload here, it shows a lot more detail about the health of the connections whilst they are in use.
1 Like
Router connected directly to test PC. Same results using 10G NIC and 1G NIC. Not a LAN problem.
OK. Will do. It’ll take several minutes for results.
Both LTE connections active. Three 1G downloads at 30 second intervals. LOTS of packet fragmentation on ATT link.
ATT only through Fusionhub. Still LOTS of packet fragmentation. Approximately the same total throughput.
Had to restart test since swapping out Verizon for ATT broke the tunnel. This is Verizon only. Some packet fragmentation. About the same total throughput. Seems like a Fusionhub problem but have no idea why.
I seem to have a lot of packet fragmentation, packet out of order, but NO packet loss on the ATT LTE connection. I tried the link on two different bands, 12 and 2, and together, but no difference.
This is a new ATT sim from MobileMustHaves, high-priced, no throttling or management. It is supposed to be totally unencumbered by packet snooping at the source. It should perform just like the highest level of postpaid plan available.
How should I interpret this?
I will have to change the outbound policy to Priority to test this - it’s Enforced now. Should I check this out?
I’d grab it just to have the comparison data, as that would show the difference where the same test is used but the traffic is being not sent via the FusionHub.
Regarding the packet fragmentation that is a bit strange, my understaning was the PepVPN should adjust MSS according to the MTU of the VPN peeers, but that may need someone from Peplink to comment further.
I guess it could be worth firing up a FusionHub in another cloud provider just to see if it is something odd with the one hosted at Vultr, I only use my own vmware environments for hosting our hubs so cannot really recommend something here although it seems quick and easy to get one running in Digital Ocean - could be worth a go just incase.
1 Like
@Jonathan_Pitts - Hi Jonathan,
Do you see anything out of order with the testing that you suggested I do?
Thanks.
@sitloongs @MartinLangmaid
Peplink Support: Please review this thread.
My router doesn’t pass the available throughput/bandwidth through to the clients.
No matter what I have done, the Speedtest (various) is stuck around 30Mbps even when I’m showing 100Mbps in the Speedfusion/router interface.
And this crazy test:
This is the speed that my throughput most represents in actual usage; browsing and streaming. There is a real problem, me thinks.
Speedfusion OFF: ATT only.
Speedfusion OFF: Verizon only.
Speedfusion ON. ATT link used via Outbound Policy.
Speedfusion ON. Verizon link used via Outbound Policy.
@sitloongs @MartinLangmaid
I reset both my Max Transit and my FusionHub to factory settings and rebuilt the FusionHub VPN. No Outbound Policies except to route traffic via FH.
Same problems.
I completely separated my LTE antennas. I’m using one outside directional, and just the paddles for the other LTE connection. Here is another example of what appears to be a limit on the bandwidth at the router. Don’t you think that there’s something wrong here?
151Mbps with SF test in router interface.
35Mbps in Speedtest at the client.
With 151Mbps, here’s what I’m getting on a 1G x 2 file download from Vultr (my Fusionhub host). Kind of sad.
File a support ticket with peplink and have them look at it, fyi I think they are on holiday till the 10th.
Could be the packet out of order issue, but not sure.
Everything looks like I’d recommend.
1 Like
Rick also said to file a ticket. I will do this in the morning.
Thanks.
@joelbean, I have checked the ticket and I saw our technical support engineer is following up with you.
Thanks.
1 Like
