FTTP jitter worse than FTTC?

I've noticed that our new Giganet FTTP connection (over CityFibre) seems to be a bit jittery in comparison with our old FTTC broadband (for example, more dropouts when playing live with the online band) so thought I'd try to quantify the difference while we have both still active. Here are two graphs over the same 24 hour period - the Giganet connection is not surprisingly getting more use but the actual bandwidth we've been using in the last 24 hours has been pretty minimal, and the random spikes have been appearing at all times of day:

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

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

As expected the base level for ping times is lower for Giganet FTTP, but there are lot of small spikes on the Giganet graph and a single dropout, whereas the Plusnet graph looks remarkably clean. Is the Giganet performance within the typical range for FTTP or should I be questioning it with them? Or is my Plusnet connection just remarkably stable?

Edited by Stargazer99 (Mon 17-Oct-22 19:03:45)

Am I assuming correctly they are both IPv4 addresses link to these graphs?

Yes, they are both IPv4 addresses. There has been a small amount of online activity in the household today but we have both been out for much of the day during which time it should have been pretty minimal. The only device that would be using the internet in the background is a single Chromecast displaying a photo carousel (one small JPEG per minute) but I'm guessing we can eliminate this because there are significant periods (up to about an hour) without a spike.

Edited by Stargazer99 (Mon 17-Oct-22 22:06:56)

What router is it?

The Giganet router is a Technicolor DGA4134, with standard factory firmware as far as I can tell. The Plusnet connection uses a Plusnet branded BT Home Hub 5 with OpenWrt firmware. We are about 400m from the Openreach cabinet and get sync rates of around 70/21, so not quite full rate but clearly using a good line.

Edited by Stargazer99 (Mon 17-Oct-22 22:12:58)

I'm no expert on BQM's but from the ones I have looked at your Plusnet FTTC graph does seem particularly good compared to others on FTTC. I'll leave others to make comment on the Giganet graph.

In reply to a post by Stargazer99:

The Giganet router is a Technicolor DGA4134, with standard factory firmware as far as I can tell. The Plusnet connection uses a Plusnet branded BT Home Hub 5 with OpenWrt firmware. We are about 400m from the Openreach cabinet and get sync rates of around 70/21, so not quite full rate but clearly using a good line.

Gut feel says it's just the router, but without swapping for another router to compare on the same connection - cant categorically prove.

To be fair the single red 'stalactite' is fine, its in the early hours of the morning and could be numerous things restarting etc. The rest of the graph is OK too, unless you're noticing actual performance impacts, I'd say nothing to worry about. You cant boast a clean dead-flat all green BQM, but its not the end of the world. 😎

Yes, I was thinking that the jitter is hardly what I'd call serious, but was wondering how it compares with the experience of others. There are one or two other Giganet/CityFibre users on this forum, and I'd be interested to hear if any of them have tried the BQM on their connection. For my rather demanding application of playing live with an online band (doesn't use much bandwidth but ping and jitter are critical) it's OK most of the time but with occasional short bursts of clicks and pops, quite likely caused by congestion somewhere along the line. For this application the Plusnet FTTC connection seems more stable.

At some point I could try using the OpenWRT modified Home Hub 5 on the Giganet connection, although I'm not sure it will have the processing power to support a full rate connection.

For your use case, I'd also check the bufferbloat, to see if its problematic whilst you're live streaming

Yes, I did the DSLreports broadband speed test, which gives A+ and the Waveform test, which gives grade A. Both have slight bufferbloat, which is almost eliminated completely using the OpenWRT router with SQM on the Plusnet connection. Apparently the DGA4134 has built-in QoS, which appears to have some effect looking at the bufferbloat test results, but this isn't configurable. However, while I was playing music online no one else was using the internet in the house, so in this case I'm not sure bufferbloat was the cause of any instability.

Edited by Stargazer99 (Mon 17-Oct-22 23:20:07)

To get a comparison when both networks are idle I turned off wi-fi on our phones overnight and unplugged the Chromecast. The results are in the attached graphics. Interestingly the jitter went up overnight on the Giganet connection - since both connections had minimal activity on them from our household I can only guess that it's either worse network congestion in the Giganet infrastructure, a minor fault with the Giganet connection or a function of the router. If anyone else using Giganet over CityFibre is reading this I would be interested to compare results.

Giganet connection
Plusnet connection

Edited by Stargazer99 (Tue 18-Oct-22 08:09:26)

Your first link is broken

Post deleted by Stargazer99

It should be working now. Just renamed the file by changing a character and edited the link - no idea why the server appeared not to like the original filename.

Giganet connection
Plusnet connection

Edited by Stargazer99 (Tue 18-Oct-22 08:12:39)

See what other Giganet on CityFibre folks have to say - but agree it’s not clean at all.

If you had another router to try it would at least rule the router in or out too.

Yes, first reaction is that what I'm seeing looks more like the result from a typical Virgin Media HFC connection than from FTTP. I might try to PM one or two other Giganet users on the forum to see if they have tried the BQM.

Your latency geting worse just after 9pm last night is a red herring, I just checked a few other BQM (on IPv4 and on other networks) and they also went unusually high latency at around the same time as yours and didn't revert until gone 9am this morning.

That's interesting. I ended up doing a hard reset of the router just before 10am as the higher latency was persisting and I realised I hadn't power cycled the router since installation 3 weeks ago. Latency appears to have dropped to the levels I was seeing yesterday before about 10pm so not fixed but reduced. [Edit - looking at the latest BQM graph closely I can see that the latency dropped back around 9.30 so just before I reset the router - so it might well be related to the generally high network latency you observed].

Edited by Stargazer99 (Tue 18-Oct-22 12:16:28)

I sent this directly via PM but perhaps more useful here...

My giganet BQM's are as follows:
BQM Giganet IPv4
BQM Giganet IPv6

For context, if you want to see BAD! My virgin business BQM (same router WAN2 but not in use unless WAN1 (giganet) falls over)
BQM Virgin Buisness
And our leased fibre line at work (100Mbps Openreach/TalkTalk backhaul I think)
My Broadband Ping

I've not noticed any problems but then I'm not using my connection in the same way as you (its running web/ftp/email/dns servers etc and then my general usage in the evenings).

I am NOT using the supplied technicolor router, but a Protcli Vp204 with a more complex OPNSense setup (multi-wan etc.). OPNsense does suffer a little with PPPoE compared to OpenWRT I think but the box is plenty powerful enough to maintain speed at 900+ (unlike my old Vigor router it replaced).

Edited by philg (Tue 18-Oct-22 16:51:43)

Thanks for that, Phil. I've just compared the IPv4 graph with my own for the same time window and they look broadly similar (perhaps mine is marginally worse but it would be difficult to claim that to an ISP and could no doubt be accounted for by variations in our usage). So maybe it's the flatness of my Plusnet graph that is remarkable rather than the spikiness of the one for Giganet

Phil has suggested that I try https://f8lure.mouselike.org/ as well - I've left a trace running overnight and the difference between that and the BQM is remarkable (see attachment, BQM trace still has frequent peaks to around 20-30ms). So I guess what I've been seeing could be the result of congestion somewhere between Giganet and the ISP that Thinkbroadband uses, which probably shows that there will be quite a bit of variation in real life.

Edited by Stargazer99 (Wed 19-Oct-22 09:26:42)