Stability options and enhanced traffic weighting

[#1]

What technically is the difference between the various stability options? Just a target SNR hike or something more clever....?

[#2]

It's not a noise margin hike.
It's basically the aggressiveness of the DLM on noiser lines.  So it could pontentially be a more aggressive noise hike if the line were unstable.
Dusty_Bin has kindly offered to try Super Stable for me

[#3]

looks like you are getting the punters so i will not have to keep my word of trialing on my line 
my dlm has settled down now it been at 12db for past 2 -3 weeks and its not tried to up, down noise margin since.
so its settled in the noise margin i had on adsl yet even adsl2 gives me abit more sync which i think is due to the  few extra down tones that are used straight after the upload tones which adsl didnt use

[#4]

Quote from: Lorian

What technically is the difference between the various stability options? Just a target SNR hike or something more clever....?

Here are my initial results:
Before (i.e. standard):

PING 212.58.254.252 (212.58.254.252): 56 data bytes
64 bytes from 212.58.254.252: icmp_seq=0 ttl=250 time=15.0 ms
64 bytes from 212.58.254.252: icmp_seq=1 ttl=250 time=10.0 ms
64 bytes from 212.58.254.252: icmp_seq=2 ttl=250 time=10.0 ms
64 bytes from 212.58.254.252: icmp_seq=3 ttl=250 time=10.0 ms
--- 212.58.254.252 ping statistics ---
4 packets transmitted, 4 packets received, 0% packet loss
round-trip min/avg/max = 10.0/11.2/15.0 ms
Status: ShowtimeRetrain Reason:	0
Channel: FAST, Upstream rate = 1016 Kbps, Downstream rate = 6615 Kbps
Link Power State: L0
Mode:			ADSL2+
Channel:		Fast
Trellis:		U:ON /D:ON
Line Status:		No Defect
Training Status:	Showtime
		Down		Up
SNR (dB):	9.4		5.8
Attn(dB):	43.0		22.8
Pwr(dBm):	20.4		12.4
Max(Kbps):	6012		20624
Rate (Kbps):	6615		1016
			G.dmt framing
K:		207(0)		128
R:		0		0
S:		1		1
D:		1		1
			ADSL2 framing
MSGc:		59		14
B:		206		127
M:		1		1
T:		1		1
R:		0		0
S:		0.9963		4.0000
L:		1662		256
D:		1		1

and after (super stable):

PING 212.58.254.252 (212.58.254.252): 56 data bytes
64 bytes from 212.58.254.252: icmp_seq=0 ttl=250 time=25.0 ms
64 bytes from 212.58.254.252: icmp_seq=1 ttl=250 time=20.0 ms
64 bytes from 212.58.254.252: icmp_seq=2 ttl=250 time=20.0 ms
64 bytes from 212.58.254.252: icmp_seq=3 ttl=250 time=20.0 ms
--- 212.58.254.252 ping statistics ---
4 packets transmitted, 4 packets received, 0% packet loss
round-trip min/avg/max = 20.0/21.2/25.0 ms
Status: ShowtimeRetrain Reason:	1
Channel: FAST, Upstream rate = 984 Kbps, Downstream rate = 6969 Kbps
Link Power State: L0
Mode:			ADSL2+
Channel:		Fast
Trellis:		U:OFF /D:ON
Line Status:		No Defect
Training Status:	Showtime
		Down		Up
SNR (dB):	11.3		5.3
Attn(dB):	43.0		22.9
Pwr(dBm):	20.5		12.6
Max(Kbps):	8436		20118
Rate (Kbps):	6969		984
			G.dmt framing
K:		218(0)		31
R:		16		14
S:		1		4
D:		32		2
			ADSL2 framing
MSGc:		59		34
B:		217		30
M:		1		4
T:		1		4
R:		16		14
S:		0.9962		4.0000
L:		1879		276
D:		32		2

That looks like interleaved to me, even though it says fast.

[#5]

Yep - That's what I thought too.
You were on auto both before and after though.

[#6]

It was looking promising, but then:
This was initiated by BT's DLM, I guess, as it happened at the same time as yesterday:
Thu, 2009-06-18 07:33:51 - LCP down.
Thu, 2009-06-18 07:33:58 - Initialize LCP.
Thu, 2009-06-18 07:33:58 - LCP is allowed to come up.
Thu, 2009-06-18 07:33:59 - CHAP authentication success
This was accidentally initiated by me:
Thu, 2009-06-18 11:14:28 - LCP down.
Thu, 2009-06-18 11:14:36 - Initialize LCP.
Thu, 2009-06-18 11:14:36 - LCP is allowed to come up.
Thu, 2009-06-18 11:14:37 - CHAP authentication success

Upstream rate = 1001 Kbps, Downstream rate = 5539 Kbps
Link Power State: L0
Mode:			ADSL2+
		Down		Up
SNR (dB):	15.3		5.6
Attn(dB):	43.0		22.9
Pwr(dBm):	20.2		12.4

So looks like I've now been bumped up to 15 dB SNR as well
I'm pretty certain that happened on the first resync today.

[#7]

Ok - Will be interesting to see if the 3 day retrain does happen.

[#8]

It was a couple of minutes late this morning, retrain happened at 07:35 UTC:

Channel: FAST, Upstream rate = 1001 Kbps, Downstream rate = 2964 Kbps
Link Power State: L0
Mode:			ADSL2+
Channel:		Fast
Trellis:		U:OFF /D:ON
Line Status:		No Defect
Training Status:	Showtime
		Down		Up
SNR (dB):	22.5		5.9
Attn(dB):	43.0		22.9
Pwr(dBm):	20.0		12.4
Max(Kbps):	9588		19984
Rate (Kbps):	2964		1001
			G.dmt framing
K:		45(0)		36
R:		16		12
S:		1		4
D:		64		2
			ADSL2 framing
MSGc:		61		29
B:		44		35
M:		1		4
T:		2		4
R:		16		12
S:		0.4803		4.5714
L:		1016		273
D:		64		2

The framing parameters did change a bit yesterday, but today it looks like the interleaving has been increased again, although ping is still at 20mS...
If someone could point me to an explanation or definition of the ADSL2 framing parameters, it would be appreciated.
I wonder if it is doing some kind of training checks at a different level each day and will then revert to something more sensible, or whether the system just reckons the previous parameters weren't aggressive enough to be consdered 'super stable'?

[#9]

Quote from: dusty_bin

If someone could point me to an explanation or definition of the ADSL2 framing parameters, it would be appreciated.

For g.dmt:
K Value - Bytes in the DMT Frame
R Value - RS check Bytes
S Value - RS code word size in DMT frame
D Value - Depth of Interleaving
For adsl2:
MSGc Value: number of bytes in overhead channel message
B Value: number of bytes in Mux Data Frame
M Value: number of Mux Data Frames in FEC Data Frame
T Value: Mux Data Frames over sync bytes
R Value: number of check bytes in FEC Data Frame
S Value: ratio of FEC over PMD Data Frame length
L Value: number of bits in PMD Data Frame
D Value: interleaver depth
And, from looking at the data, the two sets of parameters are linked as follows:
K = B+1
R = R
S = M
D = D

Quote

I wonder if it is doing some kind of training checks at a different level each day and will then revert to something more sensible, or whether the system just reckons the previous parameters weren't aggressive enough to be consdered 'super stable'?

Well, I didn't get a retrain this morning so I thought I would post a summary of the evolution over the last week since the transfer to super stable (time = n/a means that retrain didn't occur on that day):

Date:	06-16	06-17	06-18	06-19	06-20	06-21	06-22	06-23	
Time:	n/a`	07:31	07:33	07:35	07:21	07:26	07:35	n/a
Sync rate (kb/s)
Down	6615	6969	5539	2964	2964	2964	2964	2964
Up	1016	984	1001	1001	855	855	447	447
G.DMT framing
Down				
K	207	218	174	45	45	45	45	45	
R	0	16	12	16	16	16	16	16
S	1	1	1	1	1	1	1	1
D	1	32	32	64	64	64	64	64
Up				
K	128	31	36	36	43	43	16	16
R	0	14	12	12	16	16	16	16
S	1	4	4	4	2	2	4	4
D	1	2	2	2	4	4	4	4
ADSL2 framing				
Down				
MSGc	59	59	59	61	61	61	61	61
B	206	217	173	44	44	44	44	44
M	1	1	1	1	1	1	1	1
T	1	1	1	2	2	2	2	2
R	0	16	12	16	16	16	16	16
S	0.9963	0.9962	0.9993	0.4803	0.4803	0.4803	0.4803	0.4803
L	1662	1879	1489	1016	1016	1016	1016	1016
D	1	32	32	64	64	64	64	64
Up				
MSGc	14	34	29	29	19	19	64	64
B	127	30	35	35	42	42	15	15
M	1	4	4	4	2	2	4	4
T	1	4	4	4	4	4	1	1
R	0	14	12	12	16	16	16	16
S	4.0000	4.0000	4.5714	4.5714	3.2000	3.2000	4.2953	4.2953
L	256	276	273	273	255	255	149	149
D	1	2	2	2	4	4	4	4

I forgot to add that the ping has gone up from 10 mS to 25 mS last time I checked.

[#10]

Thanks.
Looking at that, the drop in speed correlates directly with the doubling of the interleaving depth.  Do you have the noise margin figures aswell?

[#11]

The last time I checked at 09:17 UTC:

		Down		Up
SNR (dB):	22.4		21.0
Attn(dB):	43.0		22.9
Pwr(dBm):	20.0		12.2
Max(Kbps):	9984		20624
Rate (Kbps):	2964		447

[#12]

Well, it definiterly looks pretty stable...
If it's really being a pain, let me know and I can move you back to stable.

[#13]

I can put up with it for a while - it's probably useful to see if it winds back over the next week or so, or whether it just stays there with silly amounts of SNR margin and error correction in place...
I do struggle to think of circumstances where users would need or want this level of stability - could be quite useful for the ISP when customers complain too much, I suppose
How are your other trialists getting on?
When I change, I'm more likely to go back to where I was on 'standard' and play with the interleaving option, although I guess we could see what 'stable' does on the way back - anyway, let's just wait and see what happens for now.

[#14]

My other trialists have been pretty quiet, so your reporting so far has been invaluable.  Although I do have one guy who had a poor connection on "standard" now reporting 11 days uptime on "stable".
Thank you very much for your help so far

[#15]

Just a couple f comments on Dusty's figures.
Before I was put on a BT banded profile I too was seeing resyncs from the BT end at just after 7.30am, often for no apparent reason. It seems to be a regular thing.
That's a big drop, from around 6900 to 2900. I was wondering whether trying the 'stable' option would give me a better speed than my current 3069, but it looks like it may end up being slower. I may still give it a try, though.
It's encouraging to see that you could get a sync of 6900 with am attenuation figure of 43. My attenuation's 39 and I've never got above 5200 (I know - it's a wiring issue).
Graham