I wonder what the expected range of the packet loss is, and whether the non-GEA connection is any better.

I wonder if 0.0033% might still be significant for some applications. 1 in 30,000 packets going missing. Doesn't sound like a lot, but could be significant.

I have some suspicion that something may be lossy somewhere, as I have noted occasional increased packet loss on TBB graphs which seems to be gateway-dependent in severity, though all gateways appear worse for me than pre-GEA for throughput and general ping latency jitter. It may not be the gateway per se, but some relationship with exactly how each route to the gateway behaves and jives with TCP congestion mechanisms.

It is always possible that if there is a gateway relationship that you get into a more working gateway just by luck of the draw when then PPPoE session is connected. You can work out which gateway you are on by looking at the first Zen hop on a traceroute to anywhere.

If our issues are related, and what I see is the current "state of the art" for GEA, then I'm really not sure how much you could expect to see with a 500/70 connection as my connection mostly tests around the 500 mark (though it's a 900), with the exception of some select routes which will test at 900 all day long.

It's possible I guess that fixes have gone in that get up to generally around the 500 mark on the GEA system, which would appear to fix your issues, but leave me falling short with my 900/115 connection.

I also note that a symptom for me is that some limited selection of routes do test at close to line rate, which of course isn't ideal, as it is always possible that the test location that Samknows uses could be on one of those good routes...

I do now have a dedicated case manager for the issue at Zen, but I've not really heard anything useful back in the last week. I'm only about 12 days off this timing out to being a month's issue from me giving notice that I intend to exit contract if they don't get it back to how it was (per their logging system against my account) so I guess the ball is in their court to resolve.

Looks like I'm going to have my very own Samknows box sent out, will be interesting to see how one behaves on my connection. Though I'm still cautious as to exactly how you can interpret the results given questions over where the test servers route to, it will nevertheless be at least another data point.

Well, my line is totally stable now latest SamKnows output I have no real idea what has changed, but I think Brandon is back on Monday, so perhaps he can shed some light. I have a nagging feeling that somewhat like FakeJake it's 'fixed itself', which is a real worry.

Out of interest, what do the other two tabs show (the Realspeed tests and the details page?).

I'm going to be interested to see what the Samknows box makes of my line, hopefully it will arrive this week and I can make some progress with this. I've done a bit of testing with thei Samknows web based test and it seems to be middle of the road on my connection at around 700Mbps, whereas Zen's own Speedtest server rarely measures that high, and some other test servers are close to line rate.

The variability between destinations since migration (and in particular the slow speed to Zen's own test server) I find particularly confusing still.

Nothing to see really! realspeed and details I don't see much difference in results between Zen's Speedtest, Speedtest.net (with Zen as a target) and Fast.com. - obviously single threaded is more variable and lower than multi-threaded tests, but multi pretty consistent here - across both of my connections at the moment.

Thanks. Have you tried running the realspeed tests - do they give you any interesting information?
I'm looking forward to receiving my Samknows box now to have a look at how it behaves on my line.

Which Fritz router are you using it with - still the 7530?

The real speed tests won't run (because of the router See here) I am using the 7530 ax on the test line, and a 7530 on my BTW line. I have seen some other comments on different forums about many other issues people are having with Zen, can't say that there's anything obviously wrong here now, but I did have a couple of months of terrible issues, and trying wacky ideas. The SamKnows box seems to indicate nothing awry with the original line GEA migrated - but there certainly was a problem, and even Zen confirmed it was present.

Hmmm.... not 'quite' all sweetness and light - Latest latency So, the latency is up by 30% .... Why? Nothing changed here, except the weather perhaps?

Maybe you're connected via a Manchester gateway - have you had any disconnections?

Was your Samknows box from Samknows themselves or from Zen? I was sent one by Zen, but they didn't actually give me any details to connect to it or the dashboard, and are connecting to it themselves to monitor. So I'm pretty much in the dark as to what it is reporting aside for the odd snippet of info forwarded by them.

In reply to a post by SteveBushell999:

Hmmm.... not 'quite' all sweetness and light - Latest latency So, the latency is up by 30% .... Why? Nothing changed here, except the weather perhaps?

Manchester gateway increases my latency up to 16/17ms from 5 to 6 via London. Spoke to Zen about it but they don't consider it to be a problem. I am manually reconnecting to London if required to get round it.