Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[GaryW]

My line was resynced by the remote end at 03:14 this morning, with INP increasing from 3.50 to 8.  I guess this was DLM in action, but according to the stats my line never went red...(based on being with BT and that supposedly being speed profile).

I'd had 3 "yellow" days - 450, 1099 and 1855 ES respectively with no SES.  There were 10 SES this morning at exactly the time of the resync, so they were possibly caused by rather than the cause of the resync, and the ES between midnight and the resync were around 40/hour.

So at no point has my line exceeded 2880 ES/day.  However, it did exceed 120 ES/hour for 5 successive hours yesterday evening.

I know that DLM works on MTBE rather than just an ES count, but I thought it was averaged over a 24 hour period.  However, the DLM intervention would imply that if it is averaged, it's over a shorter period.

Has anyone else seen similar behaviour?  Should we be ignoring the MDWS traffic lights and be looking at hourly rates?

(BTW, the INP increase has of course massively improved ES!)

[JamesK]

I had a similar issue last year when I was with BT. I think most BT customers are on the 'speed' profile with BT, and that's the assumption that MDWS makes for BT connections.

However, one day my D/S ES went over 1440, but less than 2800. The next day, to my surprise DLM took action. In a single step it increased the interleaving and banded the line.

So my guess is, in reality BT have provisioned you on the MDWS 'standard' profile instead of 'speed'.

[kitz]

Ive seen someone else fairly recently with BT have DLM action after the equivalent of using standard profile on this forum.

In the past BT supposedly used both.  They do use standard if you have IPTV.    A few years ago there was some discussion on their own forums about which profile they use and I believe someone actually managed to get their profile switched from standard to speed.  At one point I had on the DLM page as being unsure if they used speed or standard.   Then about 2yrs or so ago some said they definitely use speed.   


DLM definitely doesn't use hourly rate and I think the Speed profile number is still about right.  Any observers of my line will know that I have a daily error spike and on occasion it gets stuck, generating 1,000's of errors.   If I don't spot it soon enough, DLM will take action.   
However, there's a couple of times over the past few years when Ive exceeded 2k E/S and once when I literally scraped through by the skin of my teeth.

I think its more likely that BT are using Standard.

[Ixel]

I've never been sure if SES is just simply added to the ES for determining MTBE or whether SES may have its own threshold too. I say this because a long time ago when I was using the ASUS and trying to get it to work in a fairly stable manner I had a day where I had both ES and perhaps 100+ SES and I was under the MTBE red threshold for 'speed', however DLM still intervened next day and applied interleaving.

Any observers of my line will know that I have a daily error spike and on occasion it gets stuck, generating 1,000's of errors.   If I don't spot it soon enough, DLM will take action.   
However, there's a couple of times over the past few years when Ive exceeded 2k E/S and once when I literally scraped through by the skin of my teeth.

That's why I've developed my own bash script which is essentially my own DLM, although mine is currently configured to reduce the downstream sync rate rather than solely re-sync. If such a setting was desired I could make it ignore the banded profiles configuration in the script and solely re-sync on the state being red or purple. It has individual thresholds for ES per 24 hours, SES per 24 hours and ES per 15 minutes. The states are green (good), amber (no changes), red (bad which will change the next day) and finally purple (critical which will change immediately). It also has a caution counter setting to delay positive changes being made when the state is green, requiring consecutive days of the state being green. Unfortunately it doesn't monitor the upstream yet, as when the modem re-syncs it fetches the 'all-time' stats for the 'upstream' from the DSLAM which would trigger my script.

I haven't released it at the moment though as I don't know if some here would find it useful, otherwise I'd tidy the code a little and release it with some instructions . I mainly developed it for my current line's oddities and my preference to stay on fastpath with the lowest latency possible instead of a potentially faster downstream sync rate and higher latency. However, if G.INP arrives to ECI then the script will most likely be redundant.

[ejs]

I don't think it would make much sense to add SES to ES because by definition, a SES is also an ES. So the ES figure should already include all the seconds that are also SES.

[kitz]

I've never been sure if SES is just simply added to the ES for determining MTBE or whether SES may have its own threshold too. I say this because a long time ago when I was using the ASUS and trying to get it to work in a fairly stable manner I had a day where I had both ES and perhaps 100+ SES and I was under the MTBE red threshold for 'speed', however DLM still intervened next day and applied interleaving.

We've had doubts about SES before. However this is an exact quote from Openreach documentation on their DLM:

Mean Time Between Errors (MTBE) = Uptime / Errored Second Count. MTBE measures errored seconds only (Not HEC, CRC or FEC).

That was before G.INP and lines with G.INP may also monitor LEFTRS.

The Openreach MBTE & MTBR profile figures were last updated in May 2015 and afaik have not been amended since.
They were at one point considering monitoring over 48 hrs rather than 24hrs, and I believe it was trialled for short while,  but that idea seemed to have been knocked on the head.

[GaryW]

I've gone back through my stats for the last few months, and it looks like BT (or at least my BT line) is indeed Standard rather than Speed.  I had 1435 ES back in mid-Jan that didn't invoke DLM, then 2 days later 1840 ES that did...although at the time I thought that the 100 SES on the same day had somehow effectively bumped me over 2880 ES in the eyes of DLM.  Now that I've just seen 1855 ES (with no SES) invoke DLM again I think it's a dead cert.

I guess I'll be looking at a different traffic light on MDWS