It can't be NAT related. The router handles NAT. Nothing beyond the router even see's NAT, it cannot impact speed.

Steve - did they run any longer iperf tests? These are just short 10 second tests.

Post deleted by dect

Edited by deleted (Sat 30-Apr-22 16:53:19)

That's all the info that I have from them, but there is a crystal clear difference between the two sessions isn't there? My tests weren't iperf (although I have offered the services of my server running it to Zen).

Zen are trying another OpenReach change on Tuesday, and coming with a bunch of routers on Wednesday. Can't see how the routers will fix the behaviour that Zen have captured, but I am all ears!

Regarding 'it can't be NAT' which someone published, see the capabilities of Fortinet / Cisco / BlueCoat / Juniper / Palo Alto. They all can do deep packet inspection, and specifically throttle NAT traffic - Just Google "NAT traffic shaping". In addition, not all NAT's are equal (there are quite a few ways of bundling things up - (Wiki Network Address Translation for more details) - It could be that Apple wrap things up differently perhaps?, they DO tend to go their own way.... I am NOT saying that's the problem, merely that it could be, and clearly that it's NOT a simple problem, and to be honest nothing makes any real sense at the moment, and I an not green, I have worked in IT for over 40 years!

At least things are being done - or at least tried, which I am grateful for, and maybe, one of the approaches may guide us towards the actual problem. I remain sceptical that the GEA migration is not in some way to blame, as it was just hours after that was applied (when I got up, still unaware that anything had been done) that I noticed the glaringly obvious sudden impact on performance.

Edited by SteveBushell999 (Sat 30-Apr-22 12:44:47)

I'll be receiving a Zen Internet parcel from DPD tomorrow.

I guess this will be the whitebox.

Edit: it is the whitebox. Further tests have been done and a call between Zen and Openreach is scheduled today to try and progress further.

Edited by FakeJake (Tue 03-May-22 15:06:06)

Well....... I've just had Zen here for 4 1/2 hours, and in a nutshell:

1) They connect windows laptop over PPPOE - it's rubbish - maybe 100Mbps, but dropping packets like crazy
2) They try Technicolor router and Fritz 7300AX router, exactly the same as I see, 120Mbps, and dropping when downloading anything big
3) They confirm connecting my Mac mini over PPPOE it works fine, full speed, no drops.
4) They confirm my observations using the Apple TimeCapsule router, gives 220Mbps, and looks reasonably OK

Teleconference with Zen and Openreach, there is no conclusion, apart from saying they will push through migrating me back to pre GEA config - but WHY? What's wrong?

Essentially, the operation and speed of Mac devices is WAY different to Fritz, Technicolor, and a Windows laptop..... Go figure !!!

Well. That’s that then eh….

You just need to have a network made up exclusively of Apple computers - even for routing duties it would seem!!

What a coup for Apple. What a disaster for Zen.

In reply to a post by Pheasant:

Well. That’s that then eh….

You just need to have a network made up exclusively of Apple computers - even for routing duties it would seem!!

What a coup for Apple. What a disaster for Zen.

It does rather seem that way, and I DID consider that very possibility, but Apple internet sharing doesn't allow for port forwarding, and I need that for mail server, and Plex, and SERVARR etc etc Of course it's 'possible' but a faff, and I don't have a Mac with 2 network ports any more, and the USB3 network adapter I have doesn't support anything later than about El Capitan....

By the sound of things, OpenReach are as mystified as are Zen, but there was one little nugget..... I don't know if I heard right, but a suggestion that IPV6 works, and it's IPV4 that gets clobbered.... Perhaps I need to investigate that.....

In reply to a post by SteveBushell999:

I don't know if I heard right, but a suggestion that IPV6 works, and it's IPV4 that gets clobbered.... Perhaps I need to investigate that.....

No, that was just me trying to understand how the speedtest on the fritzbox was getting full speeds, when no other test does. So my on own fritzbox it give the url http://avm.de/service/zack-der-speedtest-fuer-ihre-b...

running that speed test gave full speeds on the test on the newer AX fritzbox that out engineer tried. But no other speed test did. When I packet captured a test on my home connection to see whether that was a UDP or TCP test (as UDP we generally can get to line rate), I saw that it was TCP, but also running IPv6. But your connection doesn't have V6, so that is not a reason for the test at your home to run full speed. Should have been obvious that it would run over V6 where its available as AVM have always been good at supporting it.

Ah, thanks for the clarity Brandon, so not worth trying to reconfigure everything here to IPV6, which would be a pain in any case! Especially as I need the IPV4 DNS entry to point to my domain anyway !

OK, well, it was a straw floating by, and asking to be clutched at !

I am as eager as anyone here to find out what will make it work, and the whole Apple versus the rest of the world is clearly interesting to say the least! Perhaps if I make my Mac dual boot via Bootcamp, so the same hardware is involved, but it runs Windows, might that give us any more info?

Edited by SteveBushell999 (Tue 03-May-22 18:05:57)