In reply to a post by c_j:

Has anyone tried playing with MTU settings to see if they make a difference?

Someone else with a long memory (BTwoolsale vs Cisco Express Forwarding?)

I think it was on pre-BT ownership of Plusnet. MTU path discovery wasn't working; reducing the MTU on the router resulted in a significantly more than doubling of throughput. It's a possible reason why different devices see vastly different speeds if their maximum MTUs are not the same.

As you say, reducing the MTU to say 1300 as a test would be very quickly done. I'm not optimistic it will work, but with so many heads being scratched, anything is worth a try.

Hi, we did run some tests changing the MTU, but with no noticeable impact.

Interestingly, had a friend over this weekend who is something of an expert on BSD TCP, and we ran some WireShark traces, which seemed to show near constant packet loss across our tests, but it was inconclusive as to where / exactly what was happening. Regarding MTU, it looked like everything was happening re: fitting into the MTU window, so don't think it's the culprit.

I'm just returned to the situation where I must wait to go back to pre GEA to see if that is the real reason, it does actually seem that all learned opinion is consistent, in that there 'seems' to be packet loss, but it's not clear where.

Zen told me on Friday that my unmigration would not be before Monday, so I'm afraid it's still a watch this space situation while BTW do their stuff.

In reply to a post by dect:

I think you're a fantasist, considering your claims of having a lease line it actually appears to be your Boss who has it.

In reply to a post by Username26:

My Boss just had a leased line put in

I suspect if he/she is actually the architect and that isn't also make believe he/she wouldn't be too happy you're taking the credit for his/her achievements.

Oddly enough it IS possible for someone to have a LL at a sat office AND a head office? you never heard of this concept before?

I'm taking 25% credit for it - seeing as it was 4 of us who did the schems. What is it with people who get so jealous they brand others names? I mean come on - if you worked harder you would also be able to afford one no?

not sure how to roll eyes on here - but you get the message

Post deleted by seb

In reply to a post by Pheasant:

Not a single thing in this thread they posted has been remotely helpful. Only serving to antagonise the OP with nonsense and fantasy to make themselves feel superior. How sad and pathetic.

Says the idiot who named themselves after a tasty bit of poultry - and is crying as my connect is better than your.

So yes you are right - how sad and pathetic. - don't go doing the lottery this week - you won't win as you seem to be totally wrong most of the time

In reply to a post by Pheasant:

...Hmmm. Had a sneaking suspicion this may turn out.

Interesting thread, my thoughts probably align with Pheasant, which is that a lot of blame is being placed on Windows versus MacOS, but it could be the make/driver of the Ethernet card. Using the same hardware to run both OSes showing no problem, but the problem showing up on routers, except the Apple Time Capsule, I wonder if all the Apple hardware is using Broadcom ethernet controllers, and the other hardware is using every other make (Realtek, intel, etc).

Well, it's not that simple, but equally it may turn out to be 'something' to do with this.

Firstly, Mac mini's I have two, 2012 and 2018, both have Broadcom ethernet, as is the chip in my 4th Gen Apple TimeCapsule. I'm not certain about the Fritz, but it looks like it may have an Atheros chip. Not sure what chip the Technicolor has.

So far so good, now the 'difficulties' of this theory:

1) The Fritz has been working fine for ages, then suddenly stops, coincident with GEA migration. Interestingly, whilst it now can only deliver around 120Mbps over the LAN interface, it also appears that referencing scope.avm.de (the AVM German based Speedtest) it gives 500Mbps down.

2) Using my Mac mini 2012 on OSX - 500Mbps down. Using same hardware running Windows 10 - 500Mbps down. Now the hard part! Using same hardware running Windows Server 2016 - 120Mbps down.

3) Not sure what Zen's support guy was using, but I saw he was only getting around 15-20 Mbps

A further note on observations today: Running FAST.COM, the default is to run 8 threads, this is in fact selectable from the more info / settings menu.... Now, setting that down to single thread gives.... Amazingly 15-20Mbps. Not sure if that is a clue perhaps, and doing Wireshark during single threaded Speedtest shows clearly many disconnections (not so many during multiple threaded test).

So it's not quite the chipset alone, but my network guro believes it has to be somewhere in the Layer 2, but that doesn't really help me.

Oh, and as an extra goodie, I tried a few VPN tests emerging in UK, Germany, Holland etc, all were in line with (or far worse in the case of long latency areas like New Zealand) to what is recorded above, so I think we can discount routing too. The inability to make sense with high latency may lead us to other areas I guess, but frankly I am wanting to just try things as they were, and see if that fixes it.

Edited by SteveBushell999 (Sun 08-May-22 22:14:50)

Update 9th May

My un-migration happened at 8am this morning, and immediately everything is back to pre 21st March performance it seems. AlI I have done is remove my Apple Time capsule, and re-connect the Fritz!Box, then a few network tweaks on the routing (they have different IP addresses, and I don't use DHCP internally for everything).

So, updated Brandon, who apparently will be looking at the wider implications this week, but it does absolutely seem that my seven weeks of carnage may now be at an end, I just hope that it doesn't magically return, and that nobody else has this experience (@FakeJake - Brandon seems aware you probably have the same issue, so I am sure he'll be in touch).

Of course this doesn't explain what actually did / didn't work right, but I am hopeful there will be some root cause analysis during the follow up.

Good luck to all the posters here, and many thanks for all the suggestions however whacky! If nothing else, it's been good to have the support you've all given.

Steve

Edited by SteveBushell999 (Mon 09-May-22 12:24:38)

As predicted, you're back to top performance.

I've already been talking to Brandon, but more so with Andrew.

I think the plan is to keep me on the Zen backhaul and work out what's causing the issues rather than migrate me back. This is probably why I have been given the Samknows whitebox, so that they can tweak and see the results without having to bother me with repeated requests for testing.

And to clarify, I'm totally fine with that. I've been credited in full for 2 months of service (so far), and although the performance is terrible Vs what I should be getting, it's perfectly acceptable for what we need (a small house with 2 people). I "survived" on 80/20 for a few years when we moved here just fine, despite 330/50 packages being available. The higher speeds were a bit of a treat "because we could" and honestly, the increased upload speed is probably more important, which hasn't wavered during this situation.

So I'm happy to carry on with my free broadband until it's properly fixed. Once it is, I'm sure I'll be able to negotiate something for myself, after all, I've provided Zen with a real life test connection for a while to their benefit.

All the best Steve, glad its finally sorted.