High packet loss on FTTC link with Pure Telecom

TomOnBoard

I got my 7/0.4 Line upgraded to 90/20 using a FTTC connection.

All is a great improvement, except that I'm getting a massive amount of packet loss on occasion, ranging from 2% to a high of over 80% for over an hour this morning. It's playing puck with online gaming efforts and some streaming activity.

I've reported it to tech support and someone is looking into it.

However, a peculiarity I'm noticing is that when there's a lot of bandwidth being used on straight file Upload/Download, packet loss appears to diminish. On the opposite end, it's at its worst when there's nothing going on.

BTW, monitoring is being done of the router (Technicolor TG588v v2) by the Broadband Quality Monitor at thinkbroadband.com. The results being obtained by this monitoring are entirely consistent with any short term monitoring I do from my network outwards to eg. 8.8.8.8.

Any thoughts as to possible reasons on the piece that is confusing me?

ED E

Forget the monitor, post your line stats.

TomOnBoard

ED E wrote: »

Forget the monitor, post your line stats.

I don't understand. Why forget the monitor? It's showing me the situation over time. I can certainly post line stats for a period of time, but I don't understand how that's going to tell anything more than is contained in the monitor.

Anyway, the substance of my confusion lies in the apparent inverse relationship between these crazy packet loss values and activity in the connection.

ED E

Once you've confirmed there's an issue we need to step back and look at the phys layer as thats what has changed. The modems diagnostic pages should have the stats, post here for review.

TomOnBoard

ED E wrote: »

Once you've confirmed there's an issue we need to step back and look at the phys layer as thats what has changed. The modems diagnostic pages should have the stats, post here for review.

There's an issue alright, and I'm leaving that to the engineers to acknowledge and resolve, so I'm not getting into the router for its stats. Im not sure what stats I could or should get in any event.

Perhaps I'm not being clear as to my reason for posting, so I'll clarify if I can.

I'm not asking for a solution to the overall problem here but simply asking what circumstances might give rise to a reduction in packet loss at the same time as an increase in connection usage?

degsie

Bad s/n ratio could be the issue, hence the stats.

ED E

Potentially a side effect of G.Vector. More frames give it a better dataset with which to correct FEXT.

TomOnBoard

I now see that I misunderstood the term line stats, as I was focussed on the statistics from the I.P. Connectivity tests, rather than the DSL or physical connection.

Here's a pdf of what I could find in the router. Hope they make sense.

TomOnBoard

Here's another page of connection info.

ED E

Modem side attenuation isnt super reliable. Are you 600m from a green street cabinet do you think? ( fibrerollout.ie ) Upstream margin in a touch low but nothing major.

3300 FEC/hr is not nothing. Along with the uncorrectable blocks.

They'll probably pull you back to 80_20 to start with but if the cab is closer than 600m which I suspect it could be then you're looking at a potential fault on your drop wire or in the house. The temporary fix in that case would be to put you on 40_15 over the weekend because a tech will be Tuesday at the earliest.

TomOnBoard

ED E wrote: »

Modem side attenuation isnt super reliable. Are you 600m from a green street cabinet do you think? ( fibrerollout.ie ) Upstream margin in a touch low but nothing major.

3300 FEC/hr is not nothing. Along with the uncorrectable blocks.

They'll probably pull you back to 80_20 to start with but if the cab is closer than 600m which I suspect it could be then you're looking at a potential fault on your drop wire or in the house. The temporary fix in that case would be to put you on 40_15 over the weekend because a tech will be Tuesday at the earliest.

Yes, the router is approx 400 M from the cabinet, allowing for bends and such, so good call..

I'm also of the view that there's a fault on the line. I'm trying to get that tested.