Traceroute is unlikely to show differences as you won't see anything happening inside the PPP tunnel, and backhaul is all L2 anyway. Unless different backhaul connects to different endpoints and you reliably know which ones serve which customers, there's not really anything you could see yourself. I think it's safe to assume that new customers would be on Zen's own network though, as presumably a large element of moving away from BTw was to reduce costs.

I feared that may be the case, but wasn't sure if anyone more knowledgable than I would have a whizzo wheeze way of telling! I completely agree that Zen would be likely to put new connections onto GEA, but let's see what the OP says re: if it's a completely new connection.

In reply to a post by jpm:

I think it's safe to assume that new customers would be on Zen's own network though, as presumably a large element of moving away from BTw was to reduce costs.

I am not sure that every telephone exchange would have Zen's backhaul present though.

Someone posted elsewhere that not only do they have to have backhaul in place, it also needs to be of a certain capacity to be able to deal with 500Mbps and 900Mbps connections; some exchanges may only have 1G links for Zen backhaul, and they apparently need to have 10G links in place to support those higher speeds. Apparently if there is only 1G link in place the 500 and 900 connections should stay on BTW.

An interesting theory, and I can only comment on MY experiences... (I have zero idea what Zen backhaul is in place!)

Pre migration on BTW solid 500Mbps....
Post migration (on the Fritz router) 20-100Mbps...
Post migration (on apple router) 200- 250Mbps...
Post migration on PPPOE with the Mac 500Mbps....
Post migration on Windows 10 PPPOE 500Mbps...
Post migration Windows server 2016 (exact same hardware as the Windows 10 above) 110Mbps...
Migrated back - 500Mbps on all of the above (except the Apple router which is ancient hardware)...

All the other hardware is modern(ish) i7 or better, all 1GBps wired ethernet, NOT WiFi, tested with minimal other traffic on the LAN (sorry never zero, email servers etc.... but repeated tests, so doubt it's wrong by much). All cables retried, replaced and proven..... all O/S updates applied etc etc.....

THIS is why I believe that it's the way that different TCP stacks handle disconnections that is the root cause of the 'apparent' differences (we KNOW there are disconnections because of the Iperf results, and TCP/UDP will always be different), I really don't think (in MY case) it's a backhaul issue at all, but as I will be getting a second line (Zen GEA) to compare with my (now) BTW line - I can make some more detailed comparisons, which will be very interesting.

Of course, my profile is only 500/70 so, your mileage may vary, again, but, that's what I have....... However, it really does seem it's NOT limited by backhaul as my PPPOE results seem to prove. There are many theories from people here, and Zen themselves, I just see the results at the moment, confusing as they are.

Edited by SteveBushell999 (Tue 19-Jul-22 22:44:04)

In reply to a post by SteveBushell999:

I do urge you to persist, I got somewhere in my case, but I am not sure that they have reverse migrated many of us (I am only aware of two cases). I do understand why Zen want to get to the bottom of this as well ! I think that fair criticism can be levelled as to the obviously less than rigorous testing (perhaps there needs to be some automatic testing carried out from the Zen end post migrations?), and also it would be helpful to tell people that 'something' may be happening.

To be fair, in the end, Zen have taken my case seriously, and I am actually keen to understand what's happened as much as anyone, hence I am trying to help them.

I hope you get a good resolution, and please keep us informed as to progress.

I am persisting for now, but my annoyance levels are rising a little. I get that there is probably a process that Zen follow though to be able to deliver customer service to thousands rather than just me, and they have to cover all the simple things first as that will fix many issues.

Yesterday they sent an Openreach engineer to check the line. He said there was no issue - very good light levels (I can almost see our exchange from my house...) and his speed test logged in via Zen on a laptop looks OK. Notably he mentioned he had been to several migrated Zen premises in the past week with similar issues - users reporting service appearing to not be delivering what they thought it should. He said nearly all cases were no fault found (one was a dirty fibre).

Also had a call from Zen support. The guy was very amiable, but scoffed at the thought that I should put any stock in the test results coming out of the Zen Ookla speedtest server (as did the prior support engineer). So I did a little test; I hired for an hour or so a 25G connected AWS EC2 Ubuntu Linux server instance in one of Amazon's datacentres and tested to the two speed test servers I was using - Zen and Swish Fibre. Both test at close to 10G down / 5G up. So the thought that these speedtest servers somehow are not capable of fast speeds is twoddle, and I'm at a loss as to why Zen's own server, which one would assume is well connected to Zen's network... is testing so much slower than some others.

Text

1 23 45 67 89 1011 1213 1415 1617 1819

Speedtest by Ookla       Server: Zen Internet - London (id = 40788)         ISP: Amazon.com    Latency:     1.70 ms   (0.06 ms jitter)    Download:  9373.96 Mbps (data used: 4.7 GB )     Upload:  4780.34 Mbps (data used: 2.2 GB ) Packet Loss:     0.0% Result URL: https://www.speedtest.net/result/c/a70cd9eb-b2e9-4dc8-a35f-29d85520dcc2     Speedtest by Ookla       Server: Swish Fibre - London (id = 34948)         ISP: Amazon.com    Latency:     2.67 ms   (0.03 ms jitter)    Download:  8420.16 Mbps (data used: 7.6 GB )     Upload:  4776.85 Mbps (data used: 2.3 GB ) Packet Loss:     0.0% Result URL: https://www.speedtest.net/result/c/1ae21155-ce84-4b65-bc16-db3a4796fca5

Yet I see big variations on these servers from my line, and they would both perform reliably at line speed prior to GEA migration. The Zen tests (and some for the other Speedtest.net servers that seem to be in a group of "similar" performance) may go below the 450Mbps level, though more often it will be around 500-550 (maybe 650 if I get lucky on a gateway) And as others have noted, the level of variation changes depending on exactly what is the PPPoE client. I also see effect from what gateway I end up on, and also which server I'm going to.

A few examples from just now on my home connection, running the speed test on the Ubiquiti router itself; though I get similar if I use a PC connected to the Fritzbox.

Text

1 23 45 67 89 1011 1213 1415 1617 1819 2021 2223 2425 2627 2829

Speedtest by Ookla       Server: Zen Internet - London (id = 40788)         ISP: Zen Internet Ltd    Latency:     3.85 ms   (0.01 ms jitter)    Download:   557.14 Mbps (data used: 440.2 MB )     Upload:   110.23 Mbps (data used: 49.7 MB ) Packet Loss:     0.0% Result URL: https://www.speedtest.net/result/c/e2e21850-d097-4976-9e9f-85283e602b08     Speedtest by Ookla       Server: Swish Fibre - London (id = 34948)         ISP: Zen Internet Ltd    Latency:     5.78 ms   (0.37 ms jitter)    Download:   774.50 Mbps (data used: 640.6 MB )     Upload:   110.27 Mbps (data used: 49.6 MB ) Packet Loss:     0.0% Result URL: https://www.speedtest.net/result/c/5a8712e3-771a-43c4-a084-f69cbdeba3c0    Speedtest by Ookla       Server: Voicehost Ltd - Norwich (id = 9060)         ISP: Zen Internet Ltd    Latency:     3.88 ms   (0.52 ms jitter)    Download:   815.37 Mbps (data used: 1.0 GB )     Upload:   110.21 Mbps (data used: 49.7 MB ) Packet Loss:     0.0% Result URL: https://www.speedtest.net/result/c/7fa14697-9c3d-4170-ac17-7b548b61caa0

Edited by jimbof (Wed 20-Jul-22 08:33:58)

In reply to a post by jimbof:

Yesterday they sent an Openreach engineer to check the line. He said there was no issue - very good light levels (I can almost see our exchange from my house...) and his speed test logged in via Zen on a laptop looks OK. Notably he mentioned he had been to several migrated Zen premises in the past week with similar issues - users reporting service appearing to not be delivering what they thought it should. He said nearly all cases were no fault found (one was a dirty fibre).

We have suspected this is not just affecting a small number of customers for some time and I think this may be further evidence to support that.

We have suspected this is not just affecting a small number of customers for some time and I think this may be further evidence to support that.

And probably why Zen are so keen as to lay in extra lines to Jake and myself! I am still convinced that there is a 'real' issue here whether it's line cards, or other faulty equipment I just don't know, but that would also seem an unlikely thing IF it is genuinely 'widespread'? The varying (sometimes sub-optimal) results across PPPOE / different user routers etc is very interesting, but probably not the issue, it's more a smoking gun I suspect.

A week to go to my second line.... Gonna be interesting! I am assuming that my existing line will be 'GEA migrated', and the new one be with BTW, but I can login to either with the new or existing credentials thereby to keep my fixed IP and domains etc.

With you and Jake assisting Zen I am surprised that haven't gone for a NDA to keep you from posting about the issues 😎

I am surprised that haven't gone for a NDA to keep you from posting about the issues

It's a good point! But, I do always try to be fair in my criticism, perhaps sometimes harsh..... eg I am dissapointed it took so long to get me back to a working connection (UnMigrated) when it was plain as day that there WAS an issue, the fact that it seems that they have rolled out an improperly tested roll-out, etc. But, I suspect that they aren't too versed in proper testing to customers rather than within their data centre (that's what the evidence suggests!). In the end however, they do seem to take it seriously now, but I am concerned that the rollout seems to continue, without any real understanding as to the 'fault(s)'. One final point, I am really disappointed as to the new Zen guaranteed minimums being only 50% of advertised headline rates, that really does seem to be a tacit admission that there is a problem to be resolved!

Perhaps Jake has signed something, he's gone quiet....

Edited by SteveBushell999 (Wed 20-Jul-22 12:11:31)