On Wed, 5 Apr 2006, poster wrote:
> Go back and look at some posts over the last month/two from Alan
> Flavell, where he explains what happened when he was on the DSL Max
> trial (he's in Glasgow on one of the exchanges which was updated
> early). He said only in the last 24 hours that the exchange tried
> to provide 7 Mbps but that wasn't stable and it caused
> disconnectsion/re-sync until it settled at a speed just below 6
> Mbps.
Just to clarify this, in case anyone mis-interprets what you wrote -
yes, I *said* it "in the last 24 hours" of your posting, *but* I was
referring to the period of 10 days following the upgrading to Max.
Once that period was over, it's been pretty stable, with only small
adjustments being observed to the line speed (I don't watch it
continuously, so I've no idea how often it's getting twiddled, I just
note slightly different values when I look at it). Maybe I'll
implement SNMP logging (the router seems to offer it), and then maybe
it would tell me about such changes, but I'm not really that fussed
about it.
> So error rates and so on can be part of the criteria. So the simple
> attenuation / SNR figures are not "magic numbers" which equate to
> any specific speed, just a range
Fair comment, but see
http://212.23.23.177/calc.htm
As a matter of interest, I tried feeding some pre-Max values into the
calculator and comparing its estimate with my actual post-Max results.
The calculator refers to "Downstream" attenuation and noise margin
figures: my router refers to "Local" and "Remote" values instead. I
presume that the values I needed to use were the "Local" ones?
Feeding in some pre-Max values of 2M, attenuation of 31.5, and SNR
margin of 20, the calculator got an estimate of 6619k.
Which is slightly optimistic, but close enough to what I'm getting
for it to be a useful prediction.
(Those numbers were just one snapshot - you might see me quoting
slightly different snapshot figures in other postings.)
> (and given the speed mentioned by Alan, with a loss of around 33db,
> mine at ~35db may easily be below his speed...
It depends also on your noise margin. As indeed that cited calculator
takes into account.
The other difference is that, previously, BT would make a conservative
choice of one fixed speed (2M, 1M, 512k...) which they considered to
be "safe" based on a set of readings. Now, the Max engine is
adjusting the speed in much smaller increments, to optimise what it
can get out of the line. It stands to reason that the results are
likely to be faster in the majority of cases - but they should also
adapt better to changing line quality, for those who suffer from such
effects.
h t h