Your GEA connection is almost the complete opposite of mine.
You have to be a little bit careful with the Slackware download as sometimes it resolves to a slower server; I was mostly using it to prove my link was good as it very often found a server that would saturate 900mbps in a single thread.

It's interesting that your TBB downloads are slower via BTW than GEA.
It's worth trying axel on the TBB download too, with one and with several threads, and seeing how that behaves. I've not used the curl method much, I only looked it up as I saw you were on a mac but figured you might not want to install axel, but thanks for installing it. Probably better using axel than curl from now on, now you have it. Note the command line to use axel to download the TBB files is a bit different as it needs to spoof a Chrome useragent, otherwise the requests get denied (probably too many people trying to make automated tests using axel).

I'm not really sure what conclusions can be drawn at the moment if any...

It's weird! Firstly, I'm fine with Brew, I use it for other things like TVHeadend etc. OK, so....

BTW

steve@Steves-Mac-mini ~ % rm bigfile ; time axel -n 1 -v -a -o bigfile http://ipv4.download.thinkbroadband.com:80/512MB.zip
Initializing download: http://ipv4.download.thinkbroadband.com:80/512MB.zip
File size: 512 Megabyte(s) (536870912 bytes)
Opening output file bigfile
Starting download

[100%] [.................................................................................................................................................] [ 28.1MB/s] [00:00]

Downloaded 512 Megabyte(s) in 18 second(s). (28801.14 KB/s)
axel -n 1 -v -a -o bigfile 0.58s user 2.34s system 15% cpu 18.233 total
steve@Steves-Mac-mini ~ % rm bigfile ; time axel -n 40 -v -a -o bigfile http://ipv4.download.thinkbroadband.com:80/512MB.zip
Initializing download: http://ipv4.download.thinkbroadband.com:80/512MB.zip
File size: 512 Megabyte(s) (536870912 bytes)
Opening output file bigfile
Starting download

Connection 27 finished
Connection 13 finished
Connection 14 finished
Connection 18 finished
Connection 11 finished
Connection 7 finished
Connection 11 finished
Connection 7 finished
Connection 11 finished
Connection 16 finished
Connection 15 finished
Connection 7 finished
Connection 13 finished
Connection 26 finished
Connection 18 finished
Connection 8 finished
Connection 1 finished
Connection 0 finished
Connection 24 finished
Connection 19 finished
Connection 15 finished
Connection 13 finished
Connection 26 finished
Connection 24 finished
Connection 19 finished
Connection 13 finished
Connection 8 finished
Connection 15 finished
Connection 0 finished
Connection 14 finished
Connection 33 finished
Connection 26 finished
Connection 22 finished
Connection 16 finished
Connection 29 finished
Connection 19 finished
Connection 1 finished
Connection 8 finished
Connection 24 finished
Connection 13 finished
Connection 12 finished
Connection 3 finished
Connection 14 finished
Connection 18 finished
Connection 28 finished
Connection 33 finished
Connection 26 finished
Connection 6 finished
Connection 37 finished
Connection 7 finished
Connection 15 finished
Connection 17 finished
Connection 23 finished
Connection 22 finished
Connection 16 finished
Connection 19 finished
Connection 29 finished
Connection 4 finished
Connection 27 finished
Connection 35 finished
Connection 8 finished
Connection 1 finished
Connection 24 finished
Connection 38 finished
Connection 39 finished
Connection 10 finished
Connection 13 finished
Connection 32 finished
Connection 9 finished
Connection 21 finished
Connection 14 finished
Connection 12 finished
Connection 36 finished
Connection 0 finished
Connection 25 finished
Connection 28 finished
Connection 30 finished
Connection 18 finished
Connection 3 finished
Connection 20 finished
Connection 34 finished
Connection 2 finished
Connection 11 finished
Connection 5 finished
Connection 31 finished

Downloaded 512 Megabyte(s) in 8 second(s). (60069.10 KB/s)
axel -n 40 -v -a -o bigfile 1.04s user 3.67s system 52% cpu 8.936 total
steve@Steves-Mac-mini ~ %

Compared to GEA:

steve@Mini-2011 ~ % rm bigfile ; time axel -n 1 -v -a -o bigfile http://ipv4.download.thinkbroadband.com:80/512MB.zip

rm: bigfile: No such file or directory
Initializing download: http://ipv4.download.thinkbroadband.com:80/512MB.zip
File size: 512 Megabyte(s) (536870912 bytes)
Opening output file bigfile
Starting download

[100%] [....................................................................................................................................................] [ 59.9MB/s] [00:00]

Downloaded 512 Megabyte(s) in 8 second(s). (61333.31 KB/s)
axel -n 1 -v -a -o bigfile 1.06s user 4.68s system 66% cpu 8.676 total
steve@Mini-2011 ~ % rm bigfile ; time axel -n 40 -v -a -o bigfile http://ipv4.download.thinkbroadband.com:80/512MB.zip

Initializing download: http://ipv4.download.thinkbroadband.com:80/512MB.zip
File size: 512 Megabyte(s) (536870912 bytes)
Opening output file bigfile
Starting download

Connection 13 finished
Connection 35 finished
Connection 26 finished
Connection 33 finished
Connection 2 finished
Connection 16 finished
Connection 24 finished
Connection 18 finished
Connection 37 finished
Connection 20 finished
Connection 27 finished
Connection 22 finished
Connection 0 finished
Connection 11 finished
Connection 26 finished
Connection 17 finished
Connection 1 finished
Connection 38 finished
Connection 2 finished
Connection 18 finished
Connection 12 finished
Connection 4 finished
Connection 19 finished
Connection 10 finished
Connection 28 finished
Connection 8 finished
Connection 16 finished
Connection 31 finished
Connection 23 finished
Connection 20 finished
Connection 27 finished
Connection 3 finished
Connection 5 finished
Connection 14 finished
Connection 9 finished
Connection 7 finished
Connection 21 finished
Connection 39 finished
Connection 30 finished
Connection 36 finished
Connection 15 finished
Connection 13 finished
Connection 34 finished
Connection 33 finished
Connection 32 finished
Connection 6 finished
Connection 29 finished
Connection 25 finished
Connection 37 finished
Connection 35 finished
Connection 24 finished
Connection 22 finished
Connection 0 finished
Connection 1 finished
Connection 11 finished
Connection 18 finished
Connection 26 finished
Connection 2 finished
Connection 12 finished
Connection 17 finished
Connection 19 finished
Connection 38 finished
Connection 8 finished
Connection 4 finished
Connection 10 finished
Connection 28 finished
Connection 16 finished
Connection 31 finished
Connection 23 finished
Connection 27 finished
Connection 20 finished
Connection 14 finished
Connection 9 finished
Connection 7 finished
Connection 3 finished
Connection 5 finished
Connection 21 finished
Connection 39 finished
Connection 36 finished
Connection 13 finished
Connection 33 finished
Connection 15 finished
Connection 32 finished
Connection 34 finished
Connection 6 finished
Connection 25 finished
Connection 29 finished
Connection 30 finished
Connection 0 finished
Connection 37 finished
Connection 1 finished
Connection 35 finished
Connection 22 finished
Connection 24 finished
Connection 11 finished
Connection 18 finished
Connection 26 finished
Connection 2 finished
Connection 12 finished
Connection 17 finished
Connection 19 finished
Connection 38 finished
Connection 8 finished
Connection 10 finished
Connection 16 finished
Connection 4 finished
Connection 28 finished
Connection 9 finished
Connection 20 finished
Connection 14 finished
Connection 7 finished
Connection 31 finished
Connection 3 finished
Connection 23 finished
Connection 27 finished
Connection 5 finished
Connection 39 finished
Connection 21 finished
Connection 36 finished
Connection 13 finished
Connection 33 finished
Connection 32 finished
Connection 15 finished
Connection 34 finished
Connection 6 finished
Connection 25 finished
Connection 29 finished
Connection 30 finished
Connection 0 finished
Connection 37 finished
Connection 1 finished
Connection 35 finished
Connection 22 finished
Connection 24 finished
Connection 11 finished
Connection 18 finished
Connection 26 finished
Connection 2 finished
Connection 12 finished
Connection 17 finished
Connection 19 finished
Connection 38 finished
Connection 8 finished
Connection 10 finished
Connection 16 finished
Connection 4 finished
Connection 28 finished
[100%] [....................................................................................................................................................] [ 55.6MB/s] [00:00]

Downloaded 512 Megabyte(s) in 9 second(s). (56902.82 KB/s)
axel -n 40 -v -a -o bigfile 1.26s user 5.70s system 67% cpu 10.375 total
steve@Mini-2011 ~ %

They are certainly 'different' - does it matter? Do I care? Aside from the latency.....
BTW:dns.google (8.8.8.8) 8.764 ms 8.166 ms 8.745 ms Versus GEA:dns.google (8.8.8.8) 11.974 ms 12.504 ms

I'm not really sure what conclusions can be drawn at the moment if any...

Damn right ! .... But why multiple Connection finished output?

Edited by SteveBushell999 (Sun 21-Aug-22 16:57:51)

In reply to a post by SteveBushell999:

It's weird! Firstly, I'm fine with Brew, I use it for other things like TVHeadend etc. OK, so....

>snip<

Damn right ! .... But why multiple Connection finished output?

Maybe the connection closed partway through and had to be re-opened for the GEA run.
It's definitely odd that our patterns are quite different. It would seem for those two tests at the moment your GEA connection is more performant than your BTW one, which fills me with a little dread as mine is due to be migrated back from GEA to BTW on 31st... Oh well.
I guess there may be local factors affecting things differently - perhaps the BTW links are worse round your neck of the woods.

BTW, I was only doing 40 connections as for me that was the level at which I got close to saturating the link with the TBB test files. It may be counter-productive in some scenarios to have such a large number of threads.

I guess at least I have a growing armoury of metrics that I can run prior to migration and post migration to evaluate how things are looking.

Well, in the 'real world', there isn't much difference between the connections that I have now, sure, I can see different numbers in the tests, but overall, the situation is chalk and cheese to the woes I had with my first GEA migration, using the Fritz router I was down to about 10-100Mbits download, and it was dropping connections all the time, but fair-enough, that must have been some (unexplained!) fault, which mysteriously has now fixed itself.

I agree, there needs to be some trepidation with any migration going by what a few people have seen, but even then (allegedly) we are in a minority, although I don't think there is any active testing of the migration, only a reliance on users complain ing, and I wonder how many would even notice - most people use WiFi all the time through six solid walls and get about 15Mbps in my experience, they don't even know what UTP is, and freak out at the prospect of using it.

Happy to run any other tests you fancy, not sure where you are that you experience different GEA / BTW connectivity, I'm in Cheshire, and it seems a good number of the people affected are in this neck of the woods.

Yes, I'm thinking think I may have jumped in the sharkpool slightly now with not much more than a steak-flavoured bathing suit in pushing for the migration back to BTW, but anyway...

Your performance did sound atrocious, mine merely a bit annoying.

I'm in Norwich City Centre. Can almost see the telephone exchange from my house.

Thanks for running the tests, if anything else occurs to me I'll let you know.

Ah yes, Norwich... I used to live in Ranworth, and later in Ditchingham near Bungay. Struggled to get 40Mb down over FTTC there! How things have changed..... Yes, when it was broken, it was terrible, I endured that for 3 months.... Let's hope your migration back to BTW goes as well as mine did.

Migrated back to BTW today.

Connection went down shortly after midnight, had to get Zen support involved to get it back working as there was some problem at their end with the radius setup so authentication was failing. Was eventually back up by 10am, by which point the family were about to do me in.

Headline news - performance is now back to pre-migration level and comfortably hitting line rates to any where that is capable it seems.
Some results

Significant minus points to Zen's billing system. After the initial email about cancellation charges and being told not to worry about it, I received a further email with an invoice attached for £££'s of pounds cancellation, and then the next day received an email telling me this amount was being taken via direct debit.

Managed to speak to someone after the Bank Holiday weekend and they cancelled that payment from being taken (apparently). The proof of that pudding will be in the eating...! We will see.

If I was being charitable I'd say the billing issues may be down to migration back to BTW not being a common case they would be having to deal with. If I was feeling a little less charitable I'd say hang about, I've spent best part of 2 and a half months here wasting time back and forth over something that seems to be an issue either with Zen's network (or how their network is implemented within their 3rd party suppliers). All resulting from a change to my connection that they pushed through that I absolutely did not request on my line.

Anyway, I'm glad it's back performing how it should, I'll be considering my position very carefully in a few months when the contract is up, as though we've got there in the end I can't help feel it was painful and didn't actually start to move along until I asked to pull the plug on the contract.

Edited by jimbof (Wed 31-Aug-22 13:15:54)

Well, at least there is a positive side there, although the billing seems odd.No really update from my end, both connections seemingly generally OK, but the GEA one does have more disconnects than I would expect I think (see here) Am watching the Zzoomm installation people at work near here, not sure how long before their symmetrical FTTP becomes active at up to 2Gbits here....

Yes, I'm glad to be back up to full speed, though a little jaded by the experience.

Have you looked at any of the other stats that the Samknows box collects? If you look on the Analytics page of the Samknows portal there are loads of other tests that it is running that might be more sensitive to differences between the connection than the multithreaded connection test that is reported as the headline number.

For instance, the difference on the Netflix downloader test is massive on mine, having gone from hovering around the 450-550Mbps mark (and being very variable) to being almost rock solid at ove 900Mbps.

I do think there is a chance they've made improvements which mean the issues are masked mostly if you "only" have a 500Mbps line. I say that as most of the tests I've run have given results in excess of 500Mbps, so it stands to reason that if that is a limitation it would only become really obvious on the lines that are contracted at the highest rate.

You can see the stepchange in behaviour from the migration in the following grabs from the Samknows Analytics, and how big the difference is between the Samknows multithreaded HTTP test and the Netflix test...
Samknows results

Here are some more pre- and post- migration traces. Migration happened on 31st August.
Most notable traces for me seem to be the Netflix download and also the RTP jitter, both of which are massively improved on BTW vs Zen GEA. Some minor differences elsewhere on latencies etc (some in Zen GEA's favour). I wonder if you can see a difference on the RTP jitter traces between the two connections?

Some More traces

My Samknows box is going back now and I'm going back to my Ubiquiti router.