Strange!

You know how everyone seems to accept that interference (and hence crc
errors) on a broadband line can increase in the evening and overnight (hours
of darkness)? Well I've been monitoring mine for a few days and find that
the opposite is occurring here. During the day with a rock steady SNR the
average crc rate is around 80 per hour, but overnight when the SNR drops
several dB and is quite variable within that dB range the average crc rate
appears to drop to between 30 and 40 per hour. Any thoughts?
--
Heard melodies are sweet, but those Unheard are sweeter
(E-Mail Removed) FN 2°°9 +, Mungo Brandybuck of Buckland

Flyiñg Ñuñ 2°10 + wrote:
> You know how everyone seems to accept that interference (and hence crc
> errors) on a broadband line can increase in the evening and overnight (hours
> of darkness)? Well I've been monitoring mine for a few days and find that
> the opposite is occurring here. During the day with a rock steady SNR the
> average crc rate is around 80 per hour, but overnight when the SNR drops
> several dB and is quite variable within that dB range the average crc rate
> appears to drop to between 30 and 40 per hour. Any thoughts?

CRC is more dependent on the type of interference. If you get e.,g.
short duration spikes - like from car or bike ignitions - that won't
knock the overall noise level up, but will spike a single packet.

And of course, you haven't said what traffic is happening.

If there is no traffic, it would not be surprising to see CRC errors
drop to almost zero., as only the keepalives and CHAP type packets would
flow.

(assuming that PPP has those..the keepalives that is)

The Natural Philosopher wrote:
> Flyiñg Ñuñ 2°10 + wrote:
>> You know how everyone seems to accept that interference (and hence
>> crc errors) on a broadband line can increase in the evening and
>> overnight (hours of darkness)? Well I've been monitoring mine for
>> a few days and find that the opposite is occurring here. During
>> the day with a rock steady SNR the average crc rate is around 80 per
>> hour, but overnight when the SNR drops several dB and is quite
>> variable within that dB range the average crc rate appears to drop
>> to between 30 and 40 per hour. Any thoughts?
>
> CRC is more dependent on the type of interference. If you get e.,g.
> short duration spikes - like from car or bike ignitions - that won't
> knock the overall noise level up, but will spike a single packet.
>
> And of course, you haven't said what traffic is happening.
>
> If there is no traffic, it would not be surprising to see CRC errors
> drop to almost zero., as only the keepalives and CHAP type packets
> would flow.
>
> (assuming that PPP has those..the keepalives that is)

Not much traffic I suppose, but the thing is that it's only been like that
for a about a week. Before then crc errors were noticeably higher in
evenings/overnight. Nothing else much has changed that I know of.
--
Heard melodies are sweet, but those Unheard are sweeter
(E-Mail Removed) FN 2°°9 +, Mungo Brandybuck of Buckland

On 15/04/10 17:55, Flyiñg Ñuñ 2°10 + wrote:

> Not much traffic I suppose, but the thing is that it's only been like that
> for a about a week. Before then crc errors were noticeably higher in
> evenings/overnight. Nothing else much has changed that I know of.

Then it seems more likely that what you're seeing is due to something
that happens locally during the day and not at night, and is only of
significance for your line.

--
<http://ale.cx/> (AIM:troffasky) ((E-Mail Removed))
18:42:19 up 6 days, 7:54, 2 users, load average: 0.11, 0.19, 0.17
It is better to have been wasted and then sober
than to never have been wasted at all

The Natural Philosopher wrote:

> If there is no traffic, it would not be surprising to see CRC errors
> drop to almost zero., as only the keepalives and CHAP type packets would
> flow.
>
> (assuming that PPP has those..the keepalives that is)

AIUI there is always data flowing - dummy atm cells.

CRCs are at a lower level (dsl frame), so I don't think that would
explain it.

alexd wrote:
> On 15/04/10 17:55, Flyiñg Ñuñ 2°10 + wrote:
>
>> Not much traffic I suppose, but the thing is that it's only been like
>> that
>> for a about a week. Before then crc errors were noticeably higher in
>> evenings/overnight. Nothing else much has changed that I know of.
>
> Then it seems more likely that what you're seeing is due to something
> that happens locally during the day and not at night, and is only of
> significance for your line.
>
Like he is actually using it by day, but not by night?

Andy Furniss wrote:
> The Natural Philosopher wrote:
>
>> If there is no traffic, it would not be surprising to see CRC errors
>> drop to almost zero., as only the keepalives and CHAP type packets would
>> flow.
>>
>> (assuming that PPP has those..the keepalives that is)
>
> AIUI there is always data flowing - dummy atm cells.
>
> CRCs are at a lower level (dsl frame), so I don't think that would
> explain it.
>

But data wont be flowing at DSL/ATM level if its not flowing above, will it?

Do those dummy ATM cells happen all the time, or as often, when there is
no actual IP data flow?

They can't really, or there would be no way you could multiplex sessions
over ATM at a contention ratio of more than 1:1!!!

alexd wrote:
> On 15/04/10 17:55, Flyiñg Ñuñ 2°10 + wrote:
>
>> Not much traffic I suppose, but the thing is that it's only been like
>> that for a about a week. Before then crc errors were noticeably
>> higher in evenings/overnight. Nothing else much has changed that I
>> know of.
>
> Then it seems more likely that what you're seeing is due to something
> that happens locally during the day and not at night, and is only of
> significance for your line.

I don't know about only my line as nothing here has changed as far as I
know. However I spotted an Openreach engineer at a neighbour's house this
afternoon. I wonder if there mind be another change soon.
--
Heard melodies are sweet, but those Unheard are sweeter
(E-Mail Removed) FN 2°°9 +, Mungo Brandybuck of Buckland

The Natural Philosopher wrote:
> alexd wrote:
>> On 15/04/10 17:55, Flyiñg Ñuñ 2°10 + wrote:
>>
>>> Not much traffic I suppose, but the thing is that it's only been like
>>> that
>>> for a about a week. Before then crc errors were noticeably higher in
>>> evenings/overnight. Nothing else much has changed that I know of.
>>
>> Then it seems more likely that what you're seeing is due to something
>> that happens locally during the day and not at night, and is only of
>> significance for your line.
>>
> Like he is actually using it by day, but not by night?

I have most traffic between 23-00 and 03-00 at the moment so I don't think
that's it. A higher crc error rate is currently occurring in the
afternoons when I'm not using my computers although my netbook in switched
on and monitoring the connection.
--
Heard melodies are sweet, but those Unheard are sweeter
(E-Mail Removed) FN 2°°9 +, Mungo Brandybuck of Buckland

Peter Crosland wrote:
>
> The SNR is the ratio between the signal and the noise whereas the CRC
> count of the number of errors so it is incorrect to compare the two
> directly. The SNR is independant of the volume of packets being
> transmitted. At night the amount of data being transmitted is likely to
> be much lower and, all other things being equal, the number of errors
> will be less as well. So the number of CRS errors is irrelevant unless
> compared to the number of packets transmitted. . To compare the number of
> errors you need to divide the number of packets by the numnber of CRC
> errors to see what percentage are being re-sent.
>
> Peter Crosland

Peter I only mentioned the SNR as that seems to be following it's usual
pattern of being steady throughout daylight hours but lower and more
variable after dusk. Today it was constant at 13 dB up to about 19-10 when
it commenced falling in stages to 9 dB around 21-30 after which it has been
varying between 9 dB and 10 dB. I have been using the computer very
sparingly in all that time - it's been mostly idle, but crc error rate,
calculated from the cumulative figures in the 2700HGV statistics was around
100 per hour for the period 12-00 'til 16-00 and around 50 per hour between
17-00 'til 23-00 during which times the only computer traffic was connection
monitoring and anti virus software automatic updating. This was on my
netbook, both my other computers were switched off.
--
Heard melodies are sweet, but those Unheard are sweeter
(E-Mail Removed) FN 2°°9 +, Mungo Brandybuck of Buckland