is there anything I can do on my end to identify any issues? The BQM did suddenly show an increase in maximum and bouts of increased avarege latency back in april while with aquiss and about 6 weeks after a swap form GPON to XGS-PON, it's exactly the same on idnet.
BQM only sends one packet per second, and can only detect 1%+ of packet loss, which is a grossly high level; it also depends on your router responding consistently to all pings, which not all do.
There's not that much you can do at your end. Well, you *could* install a perfsonar node, and also run perfsonar in a remote data centre somewhere. On the default settings, it sends 10 packets per second, so 36,000 per hour; it might see one packet drop per hour or two in your case. You can crank it up to send more packets. But it is not designed to work behind NAT; it's for network operators who have public IP addresses for everything.
And even then, all you could do is demonstrate the problem. And iDNet are likely to say "that level of packet loss is tiny, it's all working fine, go away".
Perhaps network load for their business customers mostly dictates upgrades, if peak retail load which typically occurs outside business hours can be managed by throttling single threads, something most users wouldn't notice, then it probably saves them a lot of costly upgrades that wouldn't benefit their core business.
As I said before: I think it's not network load (= congestion). I think it's packet loss. Upgrading network capacity is unlikely to make a difference. It may be a bad fibre termination that needs cleaning, or it may be a switch with insufficient buffering, or a bunch of other difficult-to-diagnose conditions.
Other ways to prove this: try a single-thread test to some servers a bit further away, say in Europe, and then further away again, say USA. If you find the single thread speed is inversely proportional to the round-trip time, that's a strong indication of packet loss.
Also, if you find the performance doesn't vary much throughout the day, then that's also a suggestion of packet loss rather than congestion - although some packet loss will be load-related (e.g. microbursts).
My last reply to support was a week ago now and I've not had a reply back. Whatever is going on I just want a resolution or to leave, don't know if I should agree to be swapped to zen backhaul, it might bring it's own problems. Paying for 2350Mbps with only 1-200Mbps actually usable doesn't seem right. It is a shared best efforts service with no guaranteed minimum but come on... that's less than 10%!
But in terms of packets they deliver, they deliver 99.998% of them.
In general it is very hard to get high throughput on TCP except on LANs where the RTT is less than 1ms.
Pages in this thread: 
Print Thread
