Dodgy FTTP BQM

Hi all,

I've got an Aquiss 330 Mbps connection via Openreach, but I'm not at all happy with the latency:
My Broadband Ping

As somebody who was with VM for 25 years I've experienced a lot worse, but the BQM you see here is (to me) pretty disappointing for an FTTP connection. Any observations, or tests I can make? Is this likely to be the upstream network Aquiss use, the local OR circuit, or something else?

I've got the option of Lit Fibre, but my first preference would be to stay with Aquiss, and my second preference would be to use another Openreach based provider.

Thanks for any thoughts on the matter

Would depend on how far you are from London, the BQM Ping Box is located in London a few miles from me, so the further you are from it will cause higher latency.

---
Paul

If the concern is the latency spikes rather than the base latency, then I would suggest applying QoS on the upstream.
That sorted out the latency spikes on my 145 Mbps FTTP.
I did not need to apply QoS on the downstream.
I applied the QoS using my own router running OpenWrt.
My old router did not have the processor "horse power" to support speeds above 80/20 with QoS.
So became my excuse to buy router with a better processor.

Good Morning,

Firstly, I don't recall if you have brought this to our attention, but if you have, can you advise what your support ticket reference is, so I can see what have been discussed so far?

The link provided appears to be a sample graph from December 3rd, nearly a week ago. Can you provide your live link as old data is not really that useful. The sample at that time, would highlight a connection being used, but the baseline latency looks fine.

What MTU are you running across your network kit? We advise 1500 end to end.

Intrigued as to why Lit would be your third choice, if you have questions about the service then I can answer them. I have had the 500Mbps package with a static IP since April and cannot rate them highly enough.

Latency can be down to lots of factors, mostly distance. Amount of hops/route too. All of which you have no control over unless you move home.

Thanks all for your attention and thoughts. Covering off a range of your ideas:

1) It's the peak latency spikes that are the concern as spiking like that illustrates an iffy connection that's poor for gaming and work video calls. On a bad day my BQM looks like a very poor Virgin Media connection where they've got an over-utilisation fault. I spent a few years as a super user in the VM's forums, I'm pretty comfortable with reading a BQM - and in their case latency faults are invariably on the local coax circuit of HFC (which shouldn't apply with OR FTTP).

2) Minimum latency is fine, my concerns have nothing to do with distance from the TBB BQM servers.

3) I suppose the router might be a problem, it's a TP-Link Deco M4, and was fine when I first connected to FTTP, and has been faultless on Virgin Media 250 Mbps connection for a couple of years before (their hub in modem mode). I ran a permanent BQM when with VM, so I'm very confident what I'm seeing at the moment is not reflecting any router limitations.

4) I believe the Deco M4 defaults to 1500 MTU

5) For reasons above, I don't think it's my router or configuration - the problem is only apparent during normal internet traffic hours, and disappears overnight. The spiking is still visible on the BQM when there's no activity on my connection.

6) I originally selected a BQM snapshot from a few days back as being representative and neither best or worst - I'm aware there's been some big games releases and sporting events this week that have created large internet traffic volumes.

7) Live BQM:
My Broadband Ping

8) For Martin, I haven't raised a service request yet because whilst I don't think it's my setup, I wanted to make sure that I wasn't missing anything, and get the wisdom of the forum.

Whats the bufferbloat of your router like? try this online tool at waveform.com

In reply to a post by AGinWorcs:

1) It's the peak latency spikes that are the concern as spiking like that illustrates an iffy connection that's poor for gaming and work video calls.

You can only conclude that with certainty if the line is unused while the tests are taking place.

If you're stuffing the pipe full of traffic (esp uploads), and have no bufferbloat protection on the outgoing side, then pings are bound to get queued up behind other packets, and that's normal.

As it happens, I have exactly the same service as you: Aquiss FTTP 300/50. I have a Mikrotik RB4011 as my router, which has a minimal amount of outbound QoS configured:

Text
1 23 4	/queue type add kind=sfq name=sfq-default sfq-perturb=10/queue simple add dst=pppoe-out2 max-limit=50M/330M name=nuc1 queue=sfq-default/default target=10.0.0.0/8,2001:4d48:XXXX:XX00::/56

Here's my BQM: IPv4 | IPv6

Of course, these aren't directly comparable to yours, because our utilization patterns could be very different. I do occasional big downloads but most of the time there's nothing other than the odd video stream (<10Mbps)

But if it turns out that the increased BQM latency you see coincides with periods of a lot of activity on your network, then it would suggest that the issue you see simply reflects queuing on your side. To see the activity on your network, of course, you'll need to get WAN interface throughput graphs from your router, for example using SNMP.

Remember the yellow line is 99th percentile, i.e. only 1 in 100 packets sees this latency. The blue line is average. And packet loss, as far as I can see, is zero. This seems pretty good to me.

"But if it turns out that the increased BQM latency you see coincides with periods of a lot of activity on your network, "

But that's the point - it doesn't. Here's a plot of a day when nobody was at home:
My Broadband Ping