Was told for interleaving to be disabled on a VDSL line i would need an engineer visit costing £

[#1]

Update

Analysis of data from the graph would seem to infer that the MBTE data held in the DLM records more or less ties up with MTBE calculations made from my linestats using the last known MBTE parameters.  
All days bar one were within 1 or 2 Err/Secs, which considering I was taking data from a graph to the nearest 50 MTBE value, is possibly about as good as you can get without actual data set numbers to be able to confirm DLM is just using Err Secs on a non G.inp line.

The only day with a noticeable discepancy was 27th Jan, where the graph shows MTBE 3500 yet my stats calculation was 1630.  There were 0 SES that day and no unusual CRCs aside from the brief daily SHINE spike.   This would infer my modem recorded more E/s than DLM did so I'm not going to worry about it as it was in my favour.  Perhaps part of the days data wasn't captured, as I notice the 25th had no data captured at all. **

--------
ETA -   Just had a thought, the fact that my modem recorded more errors than DLM -  could be attributed to the fact some data was discarded due to a Wide Area Event.

[#2]

>> 68810 es in a day.

Wow!! I just noticed that figure 😮.  I don't think I've ever seen that many before. The max is 86,400 but that would be on a permanently erroring line!
I'm not surprised the DLM took notice of your line. 2880 E/S per full 24hr period is enough to put you at ILQ red. The line must have been running like a pile of poo. 😕


Just checking if you realise too many coding violations (eg CRCs) will cause your latency to go completely to pot?

In very simplistic terms a CRC is a corrupt data packet which has to be resent and E/S is a counter of each second which contained a CRC. Tweaking your SNRM too low may well appear to increase your sync speed, but it can come at a huge cost to signal strength and noise on the line. There is no point in opening up more channel frequencies when the additional noise is going to cause dropped packets and delays.

Delays from dropped packets have a far worse effect on latency than interleaving does.
Low Interleave adds just 8ms. High Interleave adds 16ms. If you are seeing a higher increase than that, then its likely to be delay from some other source.

tldr; Do not tweak your target SNRM too low.  It will not only cause instability & errors but it will also effect your latency.

----------------------

The theory is that for interleaving to be removed, you need to keep the purple line on your graph which Anoush showed you above MTBE 300 for a certain period (days).
MTBE 300 = 280 E/Sec per full 24hr period.  

DLM needs to see your line running for a period of time with hardly any E/Secs.   Bearing in mind that the line now has Error Protection and Interleaving then the DLM will expect to see very few E/Secs anyhow.

Whilst we talk about E/Secs you must remember that DLM judges the stability of your line based on MTBE (the average mean time between E/Sec over the course of the day).    Therefore:-
High E/Secs figure = bad.
Very low MTBE = bad.

>> plugging the hub one in to the test socket and I assume I would just run speedtests, ping tests traceroutes etc to see if the connection is faster?

Linestats are the best way of checking.   The Hub One provides only limited stats, but it does show the sync speed (the speed you connect at) which is the important one.    iirc the HubOne displays this as the Data Rate.    Unfortunately I don't think it shows Err/Secs - but the good thing is that you cant tweak your SNRM down to cause additional errors  

How to get stats from the Hub One 

[#3]

I have been plugged into the test socket most of the day as some important work came up I had to do and I forgot about it. just checked my line stats and ouch!.

noticed

Mate was't able to find his plug so havent tested the hub one yet. going to undo the snr tweaks I did and plug back into the normal socket as I have noticed a bunch of issues with my connection today.  And not surprising with 446 SES .... (I havent renablled my trafic shaping and limiting again yet so could be caused by buffer bloat maybe??)

I was seeing about 15 ES an hour yesterday when plugged into normal socket.. and no ses.

I understand the basic Concepts like CRC errors etc. I have novice to intertmediate networking knowledge just in the dark as to DLM voodo . My speed seems hard locked to 61 Mbps down regardless of sync anyway. (banded?).

Sadly I have a bunch of assignments for uni due atm so my time for tinkering is limited.

PS. Your assistance so far is appreicated thanks kitz and Gandalf(sorry on a thinkpad x220 with a 1366x768 12" screen and I am not scrolling up to find your name so you shall remain Gandalf for now) .

[#4]

Back into normal socket and snr margin delta is now 0 so snr margin is now 6

DSL	Connected : Down Stream : 58570Kbps / Up Stream : 17469Kbps

I have lost about half a MB in download speed which is a shame. I shall keep an eye on error counts and see how things goes.


Downloading at 5.9-6.2MB/s atm but pf-sense is showing 1% packet loss on WAN whilst downloading.

[#5]

Apologies for tripple posting in one night but I had completey forgotten that i set up a cronjob about a year ago to create a sort of data is beautiful piece by running a speedtest on the hour every hour my machine was on.

You can see the trend of my ping get higher from the end of january.

Its not conclusive  because i am connected to a vpn for 95% of those tests. (fair warning theres over 4400 speed test results) and obviously the vpn servers could be having issues blah blah but here it is anyway.

I have no idea what I was intending to do with this data as obviously its  not very easy to graph 4400 data points.

that being said heres what a weird graph based on download speed looks like.

[#6]

Apparently this was marked as fixed maybe I accidently clicked something for the record. 100% not fixed

[#7]

I suspect that those apparently high Far End supposedly "1 day" error counts are not correct or at least not for the time period labelled. Those Far End error counts should be for the upstream, and are counted at the DSLAM and the numbers reported back to your modem. The upstream is also unaffected by you adjusting the downstream SNRM. You can't have more ES than the number of seconds elapsed. There's no interleaving/INP on the upstream. The graph of upstream MTBE was way up in the several thousands, not down in the single figures as it would have been if you actually had as many upsteam ES in 1 day as those 1 day figures state.

I suspect that those "1 day" far end counts have actually accumulated over a lot longer than 1 day.

[#8]

Yeah I believe the modem freaks out when I unplug it so one day stats become the total stats.

The day reset around 4 ISH I'll keep an eye on es. They appear much more reasonable today. Download speeds are sitting at the bare minimum of my contract requirements now though 😞

[#9]

I take one day off and look at how many posts I've got to read.

@kitz A mean time between errors (MTBE) of 86400 would indicate there are no errors on the line, because 86400 seconds is the equivalent to 24 hours, so if that's the case the line wouldn't be erroring at all.

@Youngsie97 thanks for the testing and detail. It may be worth leaving your router in the test socket for 7 to 14 days and we'll see what the data from our side looks like. I've checked the DLM profile and it's not showing your line speed to be banded: "Downstream: 0.128M-80M with Interleaving (Low). Upstream: 0.128M-20M with no error protection"

Also to remove a fix from a thread, you'd just need to go to 'Post Options' next to the post you've marked as a fix and select the option 'Not the fix' or select 'Report to Moderator' and a happy go lucky mod will be happy to remove it for you.

Good luck with the assignments though! It sounds like that's going to be more important than broadband. 

When you do get a chance to do more tinkering, feel free to post back to let us know.

[#10]

Moderator's note by Dick (Strat): Fix removed.

[#11]

I checked the post options but didn't see it my bad. Thanks strat for changing that.

My line is stable and behaving nicely at the moment the only downside is resetting the snr margin to 6 has caused me to drop 8mbps In speed (but as others point out if the decreased snr was causing errors then it wasnt worth an extra MB in download ). Buffer bloat remained a massive issue so I had to set traffic limiting to 54 Mbps to resolve pings of 200 or more whilst downloading.


To be honest until I source a power supply for the hub one there's not much I can do.

I'll be having a clear up in betweens assignments and hopefully one will turn up or maybe my mate will come through.

[#12]

No problem, if you can't find the original power supply for the Hub One, I can't recommend you use a 3rd party power adapter so I'll be happy to look at the options of sending you a new Hub One with all the associated equipment (Can't replace individual parts unfortunately) if it comes to that. Let us know how it goes. Cheers -Anoush

[#13]

Would it be possible to get  a new router sent out with a micro filter?.

Just thought that might be causing issues...

I was hoping since it was a new line the open reach engineer would install a dual socket. but alas not the case.

[#14]

Okay final post.

Could i get a router sent out assap because regardless of Qos limiting to try reduce buffer bloat the minute i start hitting my full bandwith by downloading streaming etc. my ping spikes to 260 ms + and i get booted from IRC which is really annoying as its preventing me from getting stuff done.

Hoping to rule out my equipment as the issue as quick as possible and get an engineer out to hopefully fix the issue..

I really hope my equipment does not turn out to be the issue as quite honestly having a second line behind an ISP router is useless for me and I might as well end my contract the minute i am able to.  I bought the Plusnet line  specifically to be able to use the fancy features of pf-sense including multiple gateway routing etc and the setup has been working fine until the past few months.

[#15]

Sure, I've ordered you a new router now which should land next week.

Let us know how it goes once you've got it.