G.Fast Line Attenuation stats - Only D1 showing a value

[pip75]

Hi All, 

Can anyone shed any light on what is the expected stats should be for G.Fast. 
Is it a bug in the firmware or is that normal that only D1 has a value with G.Fast
 

 

I get about 300 line rate changes once every day, does anyone else get these daily. 

 

[cookiemonsteruk]

@pip75 

Noise margins on g.fast are up the left. Speed changes are dealt with via fast rate adaptation

[pip75]

Here in text format if the images do not load. 

Broadband Link Downstream Upstream Connection Speed 149253 kbps 20727 kbps Line Attenuation D1(41.4 dB) , D2(0.0 dB) , D3(0.0 dB) U0(0.0 dB) , U1(0.0 dB) , U2(0.0 dB) Noise Margin 3.0 dB 3.3 dB

upstream noise margin is constantly fluctuating between 3.0-3.3 db

[pip75]

What do you mean by up the left?

[cookiemonsteruk]

@pip75 

They don't paint the same picture as noise margins on fttc vdsl where noise margins can indicate a line fault . The rate changes (fra) are how the line speed is changed as opposed to dlm

[cookiemonsteruk]

@pip75 

Your line is very healthy and the rate changes are perfectly normal

[pip75]

The stats may indicate a healthy line, however, my gaming experience is horrendous. 
It seems there is no real way to troubleshoot this, as the line passes the standard checks. 

A simple speed test highlights the difference between unloaded and loaded latency 
5ms unloaded and 49ms loaded, a 10 fold increase doesn't seem right.  

If the pre game matching netcode is testing for 5ms pings (generally unloaded latency) then othen once in the game the load increases your ping jumps to 50ms then you're going to have some issues with Sync.  

Out of curiousity, what do you get when you try this speed test on fast? 

[jamesn123]

@pip75 

A 'loaded' ping test is when your bandwidth is fully saturated not when your using using a couple 100kbps in a game. Your issue is likely because somebody else is using the bandwidth while you play? Remember the Sky hub doesnt have any sort of QoS so its possible to majorly affect ping by using the broadband. 

How is your gaming device connected? WiFi or ethernet? If ethernet is it directly to the hub or through powerline, switches, boosters etc?

[mae-3]

@pip75 

The G.fast stats show a healthy line and only D1 shows on the stats which is normal for G.fast because it only has 1 band both downstream and upstream because it uses time division duplexing (TDD) compared to VDSL2 which uses frequency division duplexing (FDD) and has 6 bands.

[pip75]

Hi James,  

After struggling for a few years with different providers,  I invested in a netgear gaming router and plugged it into the sky router,

This manages all WiFi and Lan connections as well as qos and bandwidth management.

I have an open NAT type, the PlayStation is wired, in the DMZ and gaming ports are prioritised.  

Its all a bit overkill just to try and get a fair connection in the game. 

I occasionally connect back to the sky box via LAN to see if it's any better but to no avail. 

Having the netgear allows me to tweak and test a few things,

Myself and others have I've found ways to improve our gaming experience and hit detection, buy lowering the MTU on the netgear to 700, the game plays better a bit better,  somehow sending smaller more frequent packets helps with my connection issue. 

I've tried is using a hybrid vpn on the router which increases ping depending on location, this however is hit and miss. 

I genuinely don't think this is a Sky issue, I had the same with Plusnet, so it makes me suspect it's openreach, profiles and the quality of the line.  

Yes I can get 150M download and 20M upload, great if it's TCP but how many duplications/retransmissions are happening and If it's UDP Packets for gaming how many are getting delayed/lost.  

My frustration is just with the next escalation point, sky say line stats are good so it just becomes a dead end. 

[mae-3]

@pip75 

Latency isnt' only caused by a fault but it is quite normal and acceptable for packet loss up to 2%, but Sky state up to 10%.

And latency is also caused by distance to the remote point and back again, eg: distance to USA would be about 100ms-150ms roundtrip and will impact games anything below about 100ms wouldn't usually be noticed.

[pip75]

@mae-3 

Thank you, I appreciate that packet loss is inevitable and subject to congestion etc. It's just the frequency of it and the impact it has that I have an issue with. 

Yes getting the highest speed achievable through vectoring, retransmission etc is great but not for time-sensitive applications. I'd happily take a lower speed over stability. 

Fundamentally there is something amiss with my connection, most people will never notice their connection is substandard, as they will only ever use the TCP protocol for browsing and email, it's only when time-critical UDP applications such a gaming and video conferencing are being used it highlights performance issues with the network. 

Here is a wired speed test example,  would you consider that jitter 50% over the base ping acceptable?

 

[mae-3]

@pip75 

Jitter is a little high on the route but would usually be the server itself at the other end. Can you try another server to test jitter to confirm that fact?

[jamesn123]

@pip75 

I would say having a router plugged into the Sky router isnt a whole lot better. You should invest in a setup that completely removes the Sky hub.

[pip75]

Hi @mae-3 

Sorry for the late reply I've been away for a few weeks. 

With the cloud flare test, unfortunately, you cannot change the server.
I've run another test today, as you can see the jitter is still high, almost 50% of the ping

Also it seems to struggle to get up to speed for upload and download, with the spikes at the start of each graph. 

Generally, I get into games around 30ms   However the in-game latency fluctuates constantly between 30-50ms.  It's not even random spikes it just seems constantly unstable. 
I'm assuming that's 20ms of jitter if the lowest ping is 30ms and the Max is 50ms?

 I've set up an iperf server on AWS to test, which I can ping at 7ms. 
Here are my results, as increase the bandwidth of the test the following happens:
* Jitter increases to around 5ms
* Packet loss starts at 10MB tests and upwards, gradually getting worse as you hit my 20MB upload limit. 
* 5ms jitter on a 7ms ping seems high 

 

Any thoughts?