FTTC dlm guide and barrier

[#1]

Fixing the crappy d-side circuits seems like a simpler solution 

[#2]

Quote from: AndyH

I would think that if DLM is going to reduce your sync speed, it will have to drop your connection no? Perhaps if it worked on the fly like ADSL, then it wouldn't be as much as an issue. Stability needs to be instantaneous (for those that want and need it), rather than delayed. I would think anyone using IPTV/Multicast would want prioritised and stable TV service.

yes, but what I am saying is, do it 'during' the instability not hours later.  Or wait for the line to already be down for other reasons (like ADSL DLM).
Also thanks Chris, for accepting my suggestions.
x47c dont confuse snrm/interleaving depth, with power masking and signal modulation.  Vectoring doesnt need control of sync capping (banding) and interleave depth to do its work.

[#3]

Quote from: BT

R.VDSL2.9 The modem shall support seamless rate adaptation (SRA) as defined in Section 13.1 of G.993.2

It shouldn't be necessary to disconnect when lowering the line sync speed because SRA is required to be supported. Although unfortunately re-connecting the PPP session is needed to update the IP profile, which would negate the possibility to avoid the disconnecting.

[#4]

ignition commented on this on tbb.
Normally SRA will flex the speed seemlesly but cannot raise the speed above the 'initial' sync speed;
So if the line speed drops via SRA, the existing IP profile is fine as it wont be throttling the line.  It may be an issue to plusnet since they use it for their traffic management but BTw side it wont be an issue.
If the sync speed increases over the initial sync speed that wont work on SRA anyway and would be a resync.

[#5]

Quote

Thanks chris Which set error rates is the dlm using for its "intelligent decision" making..... ?    {The ECI  msan's report certain error rates differently to Huawei's}

Is it similar to 21CN ?
Something along the lines of:
MTBE - Mean time between error rates (crc's per 24h   with a 15min /batch )  
With no MTBR = Mean Time Between Resyncs polled ?
^BT Handbook^

[#6]

The thresholds are slightly different on FTTC.
Unfortunately I can't advise what they are as the information isn't in the public domain but there's not much between the two.
DLM records CRC errors, FEC errors.  The DLM parameters are worked out as:
Mean Time Between Retrains (MTBR) = Uptime / Retrain Count
Mean Time Between Errors (MTBE) = Uptime / Errored+Severely Errored Second Count
MTBE and MTBR are compared against the thresholds, similar to those above.