I have just turned on my Alta Labs Route10 to use on my 10Gbe internet line in Switzerland from the provider init7. It’s an XGSPon Line so I have a Zyxel Media Converter Bridge inbetween. To connect I need to use PPPoE with Vlan Tagging (Vlan 11).
I switched from a ubiquiti Dreammachine ProMax because it’s only able to manage 3Gbps Down and 1.2Gbps upload using PPPoE and Vlan.
I was told on Reddit the Route 10 can do PPPoE at Linespeed on a 10Gig line but unfortunately I only get it to give me 5Gbps Down and 800Mbits upload speed.
Is this really the limit of the hardware or am I doing something wrong here?
Hmm, you’re definitely getting better than I was seeing, but part of it may be my modem which I haven’t yet bypassed. Unfortunately I cannot personally comment on PPPoE performance at that rate as I have don’t have the infra in place here to even offer to try and test to see what I see.
Have you tried the alternate acceleration mode? If so, any change (positive or negative)? If you haven’t, please go to the Route10 config modal, then under the Ports card, expand Advanced, then there is a menu for Acceleration.
EDIT: FWIW I was testing with a 2.5Gbps NIC against a 3Gbps symmetrical WAN (PPPoE), and seeing 1.9Gbps/2.35Gbps down/up consistent. That’s the fastest I can get at the present time, and I don’t have enough equipment here in the lab to spin something up at this time to offer to try 10Gbps
I am quite lost now, but since I also tested running openwrt on a minipc with an intel 12900k and there achieved 9Gbps down and 5Gbps up on my line I really think the issue is not really the line.
With the alternate method I get a slower download at around 4Gbps but the upload goes up to around 1.2Gbps from 800Mbit!
So there are a couple differences currently. I’m connect to a Bell GigaHub modem (Sagemcom FAST 5689E), in an IP-passthrough (pseudo-bridge-ish mode), but to copper Ethernet on the 2.5Gbps or SFP+ via THIS RJ45 to SFP+ module.
The other major difference is that while my PPPoE service requires a VLAN tag (35), but it does NOT in this pseudo-bridge mode—my modem/router is handling the tag and I’m just doing regular PPPoE over top.
So my results aren’t a direct comparison unfortunately, I was just sharing what I had been able to hit locally. I do see the same performance in whether using the copper Ethernet or SFP+ ports, but if I bypass the Route10 with say my MBP (M2 Max, 32GB) I’m seeing 2.35-2.37Gbps down/up of PPPoE performance 100% consistently. When the modem is handling the connection and the Route10 is in DHCP it has no issue maxing out that NIC either.
I do have a WAS-110 SFP+ module to bypass the GH but I haven’t yet implemented it. So admittedly my test config is a bit different. In one of the earlier firmware releases it was maxing at 1.4-1.6Gbps down, but still 2.35Gbps up, when Route10 was handling PPPoE.
Thanks. I honestly wasn’t sure if this would improve in your scenario or not, so I was a little hesitant, but I figured it also didn’t hurt to check.
One last thing, if you really don’t mind (if you do, please say so, it’s cool). In Enabled acceleration mode, if you monitor top from shell, how do the system resources look (CPU and mem)? Is it still basically a flat-line? Or does it spike when hitting 5Gbps? At 2.35Gbps I’m not even sure it registers a % point difference, but I can’t test higher than the limit of this NIC currently. This won’t fix the issue, it just tells us if it’s being properly offloaded.
Thanks a lot for checking that! Hmm, this isn’t an official statement, but just a personal comment. If the resources aren’t spiking, then that seems to indicate some issue in the chain(s) the traffic passes through, which likely indicates it’s either fixable or at least improvable. But, again, that’s a personal comment.
Okay I have another option on the table, an alltron ALL-BM410-XGSPON-GBIC Transciever which would eliminate the need for the separate Zyxel bridge and allow me to directly connect to the XGSPON Fiber of my Provider with the Route10.
It works fine on my DreamMachine, but it does not work in the Route10
Any chance of seing drivers for such a transceiver in the future? (It’s SFP+)
Can you please confirm the model of Zyxel Media Converter bridge? In a quick search, I came across this series: PM7516/PM7510-R0 Series, but I did not search if there were others. We’re interested in possibly acquiring one, but would need to verify the requirements if it’s realistic.
Thanks. That might be somewhat similar to the one I have, which specifically is the E.C.I. Networks EN-XGSFPP-OMAC-V2 XGS-PON SFP+ module. I still haven’t attempted to switch yet, so I haven’t connected it to my Route10 (yet).
I don’t think anything useful is printed to syslog, at least for this type of issue. Packet captures may help, but I don’t want to guess at what’s required… We’re intending to open a case with the chipset vendor, and I realize this is very broad but would you be willing to work with us in this context to diagnose and ideally improve multigigabit PPPoE performance? You seem very willing and eager, which I’ll thank you for in advance, but I also don’t want to assume anything if it’s not of interest.
I’m going to get at least 1x 10Gbps ordered today, and look at other changes…
As for the negotiation of the module, have you tried setting the port speed for the SFP+ interface? If not, I would try that. I would try 10Gbps first. If you open the Route10 config modal, click on the SFP+ port that it’s connected to, then where it says Speed, select 10Gbps and choose Save.
Of course, count me in! The problem I have with setting the port speed is that I’m using the cloud controller, which I can’t reach with no internet to change the port details I would use the local self hosted one but I am still afraid to pay the 50 bucks if the router is not delivering the speeds I was hoping for. When I put that sfp+ modem in I don’t get any link light whatsoever (but it gets hot, so I assume it has power).
My Zyxel Bridge is the PM7300 XGS PON Bridge.
And just to confirm again, I have 10Gbit symmetrical Internet connection
Okay, totally fair. May as well grab one+ of those free licenses while available in case you decide to switch to a container down the road. That would be an option to do this or if you’re comfortable working via shell then there should be some shell commands to run instead, which could then be reverted via shell if it didn’t work (minimizing risk with the cloud management).
Heat is a good sign of power at least. I’ve been working on a script that I think gathers useful info (although I’m iterating on it still). If you’re comfortable working via shell, I could send it directly, or if you aren’t, I could run it for you (I would need to be invited to the site).
I’ll reach out via DM about that part, but in general I intend to keep all data here for anyone following along.
Thought I should touch on this her first as I just decided to test the module I have—it will power up, heat up, but there was no link light, and I could not access it’s management page at it’s expected IP. Seems like by just plugging it in, it ends up in the same or at least similar state as yours (from a user perspective).
As soon as I set forced negotiation at 10Gbps, the link light came on, the port lit up in management, and I could ping and reach the module’s management page.
I’ll follow up via DM with a couple sets of instructions to try via shell, then we can circle back here with results after.
I would use the local self hosted one but I am still afraid to pay the 50 bucks if the router is not delivering the speeds I was hoping for. When I put that sfp+ modem in I don’t get any link light whatsoever (but it gets hot, so I assume it has power).