Virgin Media Discussion Thread

It only doubles the data AFAIK for the O2 SIM in your name, which triggers the Volt benefits.

Just checked O2 site, it actually says every O2 pay monthly plan in your name, in your household gets double data. So you could take a second O2 SIM in your name for the wife, then you'll both get double data. ;)

Thanks for that. My next question though is if I get the sim for my wife in my name will she still get O2 Priority Moments? Is that associated with the phone number or the account? Apparently the Priority moments is more important to her than the extra data :rolleyes::p
 
Haha! Good to hear from you Dave - and on a weekday, no less! :) I've been around, just not on this thread. I've been occasionally educating people (upon invitation) on Reddit/forums/chats about how to fix their 'Internet' with OpenWrt and SQM. I think that's even more fun than actually having low latency Internet under load... But I think you're patient zero for that, too!

It's currently 3.23am - I saw your post by chance after the baby woke up; so ping me again in ~12h if you've not heard from me (i.e. I went back to sleep and forgot). I don't have the originals, but I can re-run them with and without SQM for you no trouble.

I installed OpenWrt on my router and after some faffing its all up and running, also added SQM but its mostly default settings. Is there much to tweak for a VM line that could help?
 
Thanks for that. My next question though is if I get the sim for my wife in my name will she still get O2 Priority Moments? Is that associated with the phone number or the account? Apparently the Priority moments is more important to her than the extra data :rolleyes::p

Believe all lines on the account can access O2's benefits, so you'll both get priority.
 
I installed OpenWrt on my router and after some faffing its all up and running, also added SQM but its mostly default settings. Is there much to tweak for a VM line that could help?

What's "mostly default settings"? Here's my /etc/config/sqm if it helps any. Note I switched my WAN to eth0, yours may be eth1.

Code:
config queue 'eth0'
        option verbosity '5'
        option interface 'eth0'
        option debug_logging '1'
        option ingress_ecn 'ECN'
        option qdisc 'cake'
        option linklayer 'none'
        option squash_ingress '1'
        option eqdisc_opts 'docsis nat ack-filter'
        option iqdisc_opts 'docsis besteffort ingress nat'
        option qdisc_really_really_advanced '1'
        option qdisc_advanced '1'
        option egress_ecn 'NOECN'
        option squash_dscp '1'
        option enabled '1'
        option script 'piece_of_cake.qos'
        option upload '50000'
        option download '930000'

The 'docsis' tag on the ingress and egress options is the main one, but I found adding the ack-filter helped a fair bit too. You can edit the file directly, but back it up first (cp /etc/config/sqm /etc/config/sqm.original) to be safe. Then when done just save the file and issue service sqm restart. Test with flent to check results.
 
What's "mostly default settings"? Here's my /etc/config/sqm if it helps any. Note I switched my WAN to eth0, yours may be eth1.

Code:
config queue 'eth0'
        option verbosity '5'
        option interface 'eth0'
        option debug_logging '1'
        option ingress_ecn 'ECN'
        option qdisc 'cake'
        option linklayer 'none'
        option squash_ingress '1'
        option eqdisc_opts 'docsis nat ack-filter'
        option iqdisc_opts 'docsis besteffort ingress nat'
        option qdisc_really_really_advanced '1'
        option qdisc_advanced '1'
        option egress_ecn 'NOECN'
        option squash_dscp '1'
        option enabled '1'
        option script 'piece_of_cake.qos'
        option upload '50000'
        option download '930000'

The 'docsis' tag on the ingress and egress options is the main one, but I found adding the ack-filter helped a fair bit too. You can edit the file directly, but back it up first (cp /etc/config/sqm /etc/config/sqm.original) to be safe. Then when done just save the file and issue service sqm restart. Test with flent to check results.


Ah, I am a UI junkie, dont know the commands to use command line. Will try and interpret the above and see what I have, thanks. :)
 
Great ta. :)

All similar except the "dangerous" settings, will read up on those.

They're 'dangerous' because anything entered in there will be run without being checked - whether for errors (typos/validity) or 'this will set your system on fire and then delete it'. Basically they say 'This is cable, account for the packet overhead at 18 and mpu 64, and monitor/account for NAT. Don't kill the CPU - try to stay lean, monitor acks and don't let them destroy the user experience.
 
They're 'dangerous' because anything entered in there will be run without being checked - whether for errors (typos/validity) or 'this will set your system on fire and then delete it'. Basically they say 'This is cable, account for the packet overhead at 18 and mpu 64, and monitor/account for NAT. Don't kill the CPU - try to stay lean, monitor acks and don't let them destroy the user experience.

Thanks, I had a google about those settings and all seems good. Will give them a shot, though mostly happy with the defaults.

*This is all just "feels" as I have not done any testing/logging of any sort. :cry:
 
Thanks, I had a google about those settings and all seems good. Will give them a shot, though mostly happy with the defaults.

*This is all just "feels" as I have not done any testing/logging of any sort. :cry:

You'll see and feel a significant improvement with the docsis and ack-filter options. I found web pages loaded in a much snappier way and the graphs look much better on testing. Have at it!
 
BTW, @chroniclard since @Dave Taht asked me for some tests here's a sample of the Flent rrul results with and without SQM/cake. As you can see, it's well worth implementing. Notice how much more coordinated and 'together' (fairly queued) the various streams are, when SQM is enabled. I lost almost no bandwidth in the real world, and latency is literally three times less under load, on average (~22ms versus ~60ms). I tend to hit peak latency under load in the mid 20s now, versus around 80ms (or higher) without it. Marvellous!

Prepare for a (small) deluge... 56K beware (remember that? :D).

Overview without SQM:
J0Qidya.png

Overview with SQM:
XsmunAg.png

Download bandwidth plot without SQM:
I8jIvsY.png

Download bandwidth plot with SQM:
idDkH3I.png

Ping trace under load across 70 seconds without SQM:
80kFlWD.png

Ping trace under load across 70 seconds with SQM:
jkKrDwQ.png

Total bandwidth with ping overlaid without SQM:
gSPwpaM.png

Total bandwidth with ping overlaid with SQM:
3BQf3JB.png

Upload bandwidth plot without SQM:
64HFBHG.png

Upload bandwidth plot with SQM:
91D3jIe.png

Ellipsis downstream bandwidth vs latency without SQM:
tMlvCTm.png

Ellipsis downstream bandwidth vs latency with SQM:
7z2WDWN.png

Ellipsis upstream bandwidth vs latency without SQM:
wYXdsSP.png

Ellipsis upstream bandwidth vs latency with SQM:
AGeX9EC.png

Direct link to full album.

Also, speedtest.net without SQM:
Code:
$ speed

Speedtest by Ookla

Server: Vodafone UK - Manchester (id = 23968)
ISP: Virgin Media
Latency:    22.48 ms   (2.96 ms jitter)
Download:   936.46 Mbps (data used: 1.2 GB )
Upload:    50.96 Mbps (data used: 36.1 MB )
Packet Loss: Not available.
Result URL: https://www.speedtest.net/result/c/03d8afe5-71c1-451c-826d-b7b05386b14e

And speedtest.net with SQM:
Code:
$ speedy

Speedtest by Ookla

Server: Vodafone UK - Manchester (id = 23968)
ISP: Virgin Media
Latency:    22.46 ms   (0.77 ms jitter)
Download:   904.14 Mbps (data used: 1.1 GB )
Upload:    47.49 Mbps (data used: 46.5 MB )
Packet Loss: Not available.
Result URL: https://www.speedtest.net/result/c/6fb3efe9-85b0-44fb-bd5a-c26db5012b16

Check out that instant, massive, improvement in jitter! :D
 
Those look like really impressive results. How does it 'feel', can you tell when using the connection day to day?

Honestly? Night and day. I run my own off-prem DNS as you know (I think). I used to run AdGuard Home on my NAS but a couple of outages (thanks, VM) left everyone's phones/i-devices unable to use the 'net - even those who were away from home at the time. I now run it on Oracle Cloud's free tier, on a very fast ARM/Ampere instance. No matter what happens at home, DNS over HTTPS/TLS/QUIC is always available through my domain, about 20ms away in London. There's a point to this, I promise...

Without SQM, websites were always a little slow to resolve and load. Nothing serious, and everyone just got used to it, but it was always a case of click > hang on a blank page and wait a second > start loading > loaded. For gigabit Internet you'd expect that pages were basically instant, but no. Even on the Threadripper box it was the same, whether Linux, BSD or Windows. I put it down to a mix of waiting on the DNS adblocking, running privacy extensions in the browser (uBlock Origin, LocalCDN, Javascript Restrictor, DuckDuckGo, Cookie AutoDelete etc), and the huge mess that is the modern unnecessarily javascript-laden web. It just became the norm.

Now I've enabled SQM clicking links is literally just 'click/loaded'. Snappy and instant. It feels like I hugely upgraded the DNS resolution times and put in a faster CPU, even though all that actually happened was better latency under load. DNS resolution remains the same, around 20ms. Actual page load is now basically instant. Often surprisingly so.

I also notice that on Reddit, for example, the autoplay videos used to start playing and then crash quite quickly to 240p. Same for iPlayer etc. Now they load and play at 1080p and stay there. WiFi calling on our iPhones is now glitch free and clear, whereas before it would clip audio and drop syllables every few seconds. Refreshing a text feed, loading posts, scrolling on infinite scroll type sites (Reddit, Facebook (so I'm told, I refuse to use the privacy monster) and DuckDuckGo results) all just pop up instantly. Before it was 'hit the wall at the bottom of the results, stare at the blur and count to three, continue scrolling after loading'. Now they just keep on appearing as you go, even on old/slow devices. No delays.

Loads of little quality of life improvements. My wife even noticed, and commented the Internet felt much faster - what had I done? LOL That's always a good sign! That said, we also had a conversation last week where she asked me why I bothered spending so much time messing with the network, routers, Ruckus WiFi, adblockers, and Linux stuff... After all, 'everything works fine'. Yeah... that's kinda the point - you don't notice, because it all works fine and I keep it that way. If I stopped, or something broke, you'd soon notice and complain... The typical paradox for every BOFH. When you bust a nut and keep things sweet, it's thankless. The minute something goes wrong upstream, it's your fault because you're the nerd who keeps 'messing' with it. :p

Edit: A word. Derp.
 
Last edited:
Awesome. A lot of the issues you were experiencing I used to have on VM about 6 years ago. Back then 150 Mbps was the top tier (I think) yet YouTube would struggle to maintain a 1080p stream. Back on FTTP at only 40 Mbps it felt like a completely different connection as everything just felt better. It's something I'll look into if I ever end up back on VM as on the face of your experiences it seems to help (or fix) most of my common complaints about their service.
 
Awesome. A lot of the issues you were experiencing I used to have on VM about 6 years ago. Back then 150 Mbps was the top tier (I think) yet YouTube would struggle to maintain a 1080p stream. Back on FTTP at only 40 Mbps it felt like a completely different connection as everything just felt better. It's something I'll look into if I ever end up back on VM as on the face of your experiences it seems to help (or fix) most of my common complaints about their service.

It’s certainly a vast improvement, but I’d switch to proper FTTP in a heartbeat. OpenReach reckon my street will have it installed between this spring 2022 and spring 2025. Every area up to our boundary is already live, so fingers crossed it’s towards the former rather than the latter. Time to spec out a 10Gb LAN and WAN. :D
 
It’s certainly a vast improvement, but I’d switch to proper FTTP in a heartbeat. OpenReach reckon my street will have it installed between this spring 2022 and spring 2025. Every area up to our boundary is already live, so fingers crossed it’s towards the former rather than the latter. Time to spec out a 10Gb LAN and WAN. :D
I have the luxury of both and work pay towards for one connection which I have with Zen (900). I'd be lying if the thought hadn't crossed my mind of getting VM broadband only (or even as part of a deal when Sky TV is up for renewal) and toying around with dual WAN.
 
I have the luxury of both and work pay towards for one connection which I have with Zen (900). I'd be lying if the thought hadn't crossed my mind of getting VM broadband only (or even as part of a deal when Sky TV is up for renewal) and toying around with dual WAN.

That'd certainly give you some interesting scope for load balancing and round robin. Prioritise latency sensitive stuff over FTTP and throw large downloads onto the VM pipe... Maybe one with VPN and one without. Choices, choices. :D
 
While I just tremendously enjoyed your enthusiasm, and I look forward to more data @croniclard ? Flent's comparison plots can be easier on the eyes. (Data->add other files), and you can also scale the plot to the same scale as another.
 
While I just tremendously enjoyed your enthusiasm, and I look forward to more data @croniclard ? Flent's comparison plots can be easier on the eyes. (Data->add other files), and you can also scale the plot to the same scale as another.

Ohhh... Now you tell me. :p I admit I didn't know that. Off to play (but won't post any more haha)!

Edit: Oh, yes. That's much better!
 
Back
Top Bottom