In reply to a post by andrewhearn:

Yes, we re-enabled our LINX session to NetConnex who host the BQM hardware - though, I'm not really sure why it makes such a difference as neither us or NetConnex see a problem on LINX generally!

Is it something you're going to look into? I run a BQM against z.witless as well as against my own connection, as doing so allows me to see what is me and what is further up the chain. These were the BQMs at the point it changed:
AAISP z.witless BQM
My home AAISP FTTP900 BQM
You can see it affects BQMs to z.witless, and not just to end user endpoints in homes.
Whether there is any customer impact, who knows, but maybe it shows all isn't rosy along that route for some reason?

Maybe the thinkbroadband staff can enlighten us as to whether this affects other ISPs, or just AAISP? It seems unnusual, as most of the other BQMs I've seen look generally clean, except for periods when you might expect the ISP is having congestion, or the user is hammering their connection, etc. This seems odd in that there is a permanent noise floor overlaid on what would otherwise be the exemplary performance you're buying ito with AAISP.

On my side, it annoys me a little as I use the BQMs as a way of trying to figure out if I need to check anything over on my network, and now I basically have to mentaly block the noise and compare it to the LNS BQM to decide all is ok. It's not a big deal as the connection is basically performant, but just an annoyance.

Connected my edge router to the test- LNS last night and noticed that the BQM pings to the single IPV4 address I got with the connection much improved, however the BQMs to an IP address in the block of 8 I additionally have from AAISP, which terminate at a machine beyond the edge router are still looking a bit "hairy".

See here from 10pm onwards when I switched to the test LNS:
Edge router

Internal router

@andrewhearn - is this something you'll look into? It irks me that my parent's NowTV connection looks cleaner than this!
Parent's NowTV

It seems that IPv6 is going over the better routing again (lonap) as mine went back to looking flat on IPv6. My IPv4 is still going via Linx and looked like yours, but then improved a bit, not sure why as I hadn't changed anything this end, it still has spikes of higher latency but no where near as bad as it was, although not as good as it had been via Lonap.

Perhaps raise a ticket with AAISP?

I just made an IPv4 routing change (2023-06-29 12:50) - See how that looks...

@andrewhearn Just some feedback, for me my IPv4 still has spikes and doesn't seem to have changed with the routing switch (I do see the new routing, added below), but IPv4 was never looking as rough as @jimbof graphs on the recent change back to Lonap. IPv6 looks like a nice day out on calm waters via Linx. Incidentally I checked usage in my firewall and around 70% of traffic is IPv6 in/out, so the rougher looking IPv4 isn't due to me using the connection as far as I can tell, plus there are spikes when I'm tucked up asleep. Before IPv4 switched back to lonap, so when previously going via Linx, the BQM on IPv4 was flat with nothing like the spikes seen now.

Perhaps the routing from TBB pinger to my IP is stuck going over the Lonap route, and is only picking up the Linx routing when I trace the route out, not sure if that can happen? Anyway my BQMs are in the signature for checking.

Text

1 23 45 67 8

1    <1 ms    <1 ms    <1 ms  pfSense.localdomain [192.168.1.1]   2     7 ms     7 ms     7 ms  x.witless.thn.aa.net.uk [90.155.53.133]  3     8 ms     7 ms     7 ms  k-aimless.thn.aa.net.uk [90.155.53.101]   4     8 ms     8 ms     8 ms  linx-gw1.thn.ncuk.net [195.66.224.240]  5     8 ms     8 ms     8 ms  ae11-11.edge-rt2.thdo.ncuk.net [80.249.97.21]   6     8 ms     8 ms     8 ms  te1-51-36.core-rs3.thdo.ncuk.net [80.249.97.72]  7     8 ms     8 ms     8 ms  po5-32.core-rs4.thdo.ncuk.net [80.249.97.90]   8     8 ms     8 ms     8 ms  pingbox1.thinkbroadband.com [80.249.99.164]

Edited by E300 (Thu 29-Jun-23 17:04:50)

You are correct, the route from TBB to A&A would have been via LINX still. See how it is from now, as it should now be via LONAP...

I'm seeing similar spikiness on IPv4 whereas my IPv6 graph is pretty flat:

IPv6
IPv4

Thanks for the reply, I will see how it looks.

To recap, for me the IPv4 was never as spiky as jimbof BQMs, it had been in the past, then the routing got changed and it was really flat, then the recent route change which prompted the lastest posts from jimbof and their BQM seemed to have a full regression in latency spikes, whereas my IPv4 BQM started getting more spikes but not a full regression. IPv6 has remained flat. Unfortunately jimbof doesn't seem to have live BQMs in their signature and they haven't come back yet to say if it has improved.

I also see poppadum has posted a much worse IPv4 BQM (that's what I refer to as a full regression which I haven't seen this time around). Odd how my IPv4 isn't showing the exact same behavior. If you want me to do any tests or trace routes let me know.

It doesn't seem to have improved as yet.
There is a BQM which is running directly against z.witless which exhibits it, too.
With a sprinkling of packet loss, too, it would seem...
My Broadband Ping

I wonder if the issue is related to z.witless? I'm on x.witless so set up a BQM on that directly, as my IPv4 is pretty much okay, nothing like yours anyway.

It is early minutes so not showing much as I type, however its below for anyone wanting to take a gander.

https://www.thinkbroadband.com/broadband/monitoring/...

Edited by E300 (Fri 30-Jun-23 16:06:09)