Kitz ADSL Broadband Information
adsl spacer  
Support this site
Home Broadband ISPs Tech Routers Wiki Forum
 
     
   Compare ISP   Rate your ISP
 
Please login or register.

Login with username, password and session length
Advanced search  

News:

Author Topic: Interesting DLM Behaviour  (Read 1202 times)

bkehoe

  • Just arrived
  • *
  • Posts: 15
Interesting DLM Behaviour
« on: January 11, 2018, 05:47:28 PM »

I just thought some might be interested in an observation with how DLM has been treating my connection (I'm bkehoe on MDWS).

I moved address back in July from a location with a Huawei cabinet and had GINP with my Zyxel 8924. Upon moving in here it took almost 2 months before GINP was enabled and made a substantial increase in sync rates as expected having had an interleaved connection up till then. I'm in a relatively densely populated part of Glasgow with partial underground and overhead cables and a Huawei cabinet about 100m further away than the PCP (conservation area). No extra internal wiring/extensions, feed from the pole (which does pass directly overhead an electrical substation) joins to an internal CAT5 terminating with a Mk4 filter in a utility room (previously Mk3 but no noticeable difference) and 1ft CAT5 cable connecting to my 8924. Presuming the underground element of the cable takes shortest route to PCP cabinet, I estimate a distance of no more than 250m to the Huawei cabinet.

Anyways, GINP was active on downstream only and theres always been a steady amount of ES on the upstream but 0 on downstream. Upstream usually in 'amber' range on MDWS but on occasion there seems to be a crosstalker/external noise source that somewhat impacts my downstream SNR and increases the upstream ES. A few months ago this happened and had enough of an effect to give me GINP on upstream as well. Thereafter 0 ES up and down. Surprise surprise, 2 days later DLM intervenes and changes my target downstream SNR to 5dB :cool:. However it also removes upstream GINP so the upstream ES come back and remain in the amber range. No further movement from DLM on target downstream SNR was to happen for many weeks.
A few days ago the crosstalker (or whatever the source of upstream ES is) reappeared and barely pushed my upstream ES to red. DLM intervened and put GINP on upstream again...good. Now zero ES up and down again.
Today, 2 days later DLM changes the target SNR to 4dB and leaves the upstream GINP enabled so for the first time in 6 months I'm now syncing at 80MBit down (albeit with a lower IP profile, of about 73Mbit due retx high - REIN enabled).

So to cut a long story short it would appear to me that DLM is allowing an amber condition on the upstream to inhibit improvements to the downstream sync that it can safely do and only due to my 'luck' of getting occasional upstream GINP enabled did I finally get to the full capacity of my line. Maybe others have seen this behaviour before but I haven't seen anyone mention it here.
Logged

tiffy

  • Reg Member
  • ***
  • Posts: 297
Re: Interesting DLM Behaviour
« Reply #1 on: January 11, 2018, 08:06:13 PM »

Very interesting observations especially as it's quite unusual to see US G.Inp applied, very rarely reported.

I note from your MDWS ILQ indicator (appreciate that strictly speaking it isn't), with BTw showing as your ISP you should be on a "speed profile" and never actually achieved ILQ "red" state, however, it was obviously sufficient to prompt intervention from DLM and US G.Inp application.

I would agree, you appear to be on retx high profile with B0 INP Rein at "1" since 7/10/17, from my line experience I would have expected DS INP to have been > 50 with retx high profile as has always been the case with my line when changing from low to high, usually goes from 47 to 51/52, although appreciate that all lines are different.
Can't comment on the US INP parameters as I have never seen a line with US G.Inp in detail.

It's amazing the difference the G.Inp makes to the error rates, even with the penalty of lower data throughput ratio associated with retx high profile which now appears to be more common than not, in my opinion, a small price to pay for a more stable connection.

Hope your line manages to retain US G.Inp, will be interesting to follow your MDWS trends and see what plans DLM has for your connection in the longer term.
Logged
Plusnet 40/10 - DSLStats/RPi - ZyXEL VMG1312-B10A - Huawei Cab - MDWS (sadly gone 12/03/18)

bkehoe

  • Just arrived
  • *
  • Posts: 15
Re: Interesting DLM Behaviour
« Reply #2 on: January 11, 2018, 08:30:12 PM »

Should have mentioned, yes I'm on BT Infinity 2 so Speed profile.

I think before the DLM intervention a few days ago (So on 5dB target downstream SNR and G.INP just enabled on downstream) my INP value was 51 and I was definitely retx high.

Todays intervention with INP values of 45 down, 47 up while retaining the REIN and retx high (I checked IP Profile on BT speediest) is quite different.

TBH the u/s GINP or not never affected the u/s sync speed (it always synced at 20MBit) so even if GINP gets removed from the upstream presumably based on the DLM behaviour so far my d/s will stay the same. However if I could get retx low that would be nice!!
Logged

vic0239

  • Reg Member
  • ***
  • Posts: 320
Re: Interesting DLM Behaviour
« Reply #3 on: January 12, 2018, 08:14:12 AM »

I had a resync this morning and G.INP has been reapplied to my US as well.
Logged

tiffy

  • Reg Member
  • ***
  • Posts: 297
Re: Interesting DLM Behaviour
« Reply #4 on: January 12, 2018, 08:50:42 PM »

@ vic0239

Yes, you certainly had a very bad US ES day yesterday, 11th, with daily total of 5400 you were well into ILQ red even on speed profile, DLM intervention was fairly predictable.
Still, good to know the system works and I can't help but to be amazed at the difference G.Inp makes, currently running at approx. 25 US ES/day !
Hope the US data throughput has not suffered too much, I assume that retx high/low profiling applies to US as well, wonder if this is applied independantly of DS retx profile or must they both be the same ?
I have no idea as never had US G.Inp active to date.

@ bkehoe

Quote
Todays intervention with INP values of 45 down, 47 up while retaining the REIN and retx high (I checked IP Profile on BT speediest) is quite different.

My observations of US INP being > 50 when on retx high profile is purely from my experience with my line although I have seen this mentioned a few times in various posts by other members, however, it appears not to be consistant and the most definitive indicator (certainly of DS retx profile) is B0 INP REIN being "1" for high and "0" for low retx profile.
Of course it's easy to calculate the data throughput ratio from synch speed and BTw BRAS profile which will give a definitive indication of retx high/low profile. (provided your BRAS profile is correct)
Can't see how this method would be applied for US data throughput ratio calculation as no US BRAS profile.

Quote
TBH the u/s GINP or not never affected the u/s sync speed (it always synced at 20MBit) so even if GINP gets removed from the upstream presumably based on the DLM behaviour so far my d/s will stay the same. However if I could get retx low that would be nice!!

I can only comment on DS G.Inp to date.
I'am on a PN 40/10 package, was 40/2 until the free upgrade by PN last November.
DLM has changed my DS retx profile a number of times, have always found that my DS synch rate would increase on retx high but with the associated decrease in data throughput ratio my data speed remained about the same, the inverse applied when on retx low profile again leaving my data throughput speed virtually the same.

My last excursion to DS retx low profile occured when my package was upgraded from US 2 to 10 Mbps, re-synch'ed on low profile (had been running on retx high previously) however, after a few days, DLM put me back on retx high even though my DS ES/CRC error rates were consistantly in single figures, line very stable !
It appears to be very difficult to convince DLM to change back to retx low profile even on a very stable and error free line.

Will be interesting to see if DLM will change your line to US retx low profile in the longer term.   
Logged
Plusnet 40/10 - DSLStats/RPi - ZyXEL VMG1312-B10A - Huawei Cab - MDWS (sadly gone 12/03/18)

vic0239

  • Reg Member
  • ***
  • Posts: 320
Re: Interesting DLM Behaviour
« Reply #5 on: January 22, 2018, 04:56:12 PM »

Resync at lunchtime today, G.INP removed from US.  :no:
Logged

tiffy

  • Reg Member
  • ***
  • Posts: 297
Re: Interesting DLM Behaviour
« Reply #6 on: January 22, 2018, 08:23:33 PM »

Note, you are currently off line on MDWS, since 19:01 hrs.

Yes, can't see any reason for your DLM RDI re-synch other to facilitate the removal of US G.Inp.
Will be interesting to see if your line reverts to it's US pre-G.Inp performance of 11th with respect to ES rate and how DLM reacts to that.

As it happens, I also had an unexpected DLM RDI re-synch late afternoon I believe the purpose being to change my DS retx profile from high to low, have gained approx 1.6 Mbps on both DS synch and data throughput speeds, BRAS I/P profile increasing by the same amount also.
Correspondingly, my DS data throughput ratio has increased from 92.6 to 96.6% as would be expected with retx high to low profile change.
Not a drastic change but significant enough on a 40/10 contract and certainly a step in the right direction.
My line has been on retx high profile since 20/11/17 but never really believed that this was merited as always ILQ green and very low on error rates.

Don't know if there is any connection but I have been experimenting over the last 2 weeks with modem synch limiting have found that imposing a very small limit on US speed produces quite a marked change in US ES rate, not that it ever was excessively high, just higher than it used to be.
I have set the DS modem limit to max for my line, 40 Mbps as I don't need to limit this and was not sure if the Telnet command format had to include a DS limit figure as well as a US figure.

Results can be observed on MDWS since 10/01/18, US ES rates.
« Last Edit: January 23, 2018, 03:11:44 PM by tiffy »
Logged
Plusnet 40/10 - DSLStats/RPi - ZyXEL VMG1312-B10A - Huawei Cab - MDWS (sadly gone 12/03/18)

NewtronStar

  • Kitizen
  • ****
  • Posts: 4811
Re: Interesting DLM Behaviour
« Reply #7 on: January 22, 2018, 08:45:16 PM »

I have been experimenting over the last 2 weeks with modem synch limiting have found that imposing a very small limit on US speed produces quite a marked change in US ES rate, not that it ever was excessively high, just higher than it used to be.

May even try that myself but your Circuit is very quiet hardly any RFI & REIN to be seen very low FEC's and shine events very rare  ;)
Logged

tiffy

  • Reg Member
  • ***
  • Posts: 297
Re: Interesting DLM Behaviour
« Reply #8 on: January 23, 2018, 03:09:17 PM »

Since the DLM RDI re-synch yesterday, even though my US 5,500 Kbps modem synch limit is still in place my US ES rate has increased again to date, however, as you have said, my line is still very low noise and nowhere near US ILQ amber even, very unlikely to ever cause an issue.

The modem limiting exercise was just an experiment really to see if it worked and what effect it would have on US ES rate.

Perhaps, we are back to the points made in the first post of this thread regarding interaction between US & DS synch/data throughput rates, in my case with yesterday's DLM re-synch my US & DS synch rates remained exactly the same (to be expected with the modem limits still being active) however, my DS data throughput rate has increased by approx. 1.6 Mbps due to DS retx profile change from high to low.
Logged
Plusnet 40/10 - DSLStats/RPi - ZyXEL VMG1312-B10A - Huawei Cab - MDWS (sadly gone 12/03/18)

vic0239

  • Reg Member
  • ***
  • Posts: 320
Re: Interesting DLM Behaviour
« Reply #9 on: January 23, 2018, 05:20:52 PM »

My US SNRM has been all over the place recently. I think I have a new disturber who's influence is only on US. Over the Christmas period it was particularly noticeable (a neighbour thought it a hoot to bedeck a 15 foot tree in their garden!). While G.INP was active this last week ES were negligible. Why not just leave it active?  ???
Logged

tiffy

  • Reg Member
  • ***
  • Posts: 297
Re: Interesting DLM Behaviour
« Reply #10 on: January 23, 2018, 08:06:15 PM »

Quote
While G.INP was active this last week ES were negligible. Why not just leave it active?  ???

Yes, why indeed, the wonderful logic of BTw DLM !

I know quite a few forum contributors are not big fans of DLM, I'am of the opinion that some form of dynamic line management is essential, while certainly not claiming to ever have the technical knowledge to fully understand the workings of such a system, I just wish there was a lot more clarity on the operation & expectations of DLM, even the many more knowledgable forum members here seem to struggle due to lack of concise operation details.
Logged
Plusnet 40/10 - DSLStats/RPi - ZyXEL VMG1312-B10A - Huawei Cab - MDWS (sadly gone 12/03/18)