[kwlug-disc] Hitting Bell / Rogers where it hurts?

[Eric Gerlach]

On Thu, Feb 3, 2011 at 10:08 AM,  <unsolicited at swiz.ca> wrote:
> Note to list managers: I did not see this message from Eric come across.
> Perhaps a glitch on my end, but I don't see it.
>
>> On 01/02/11 05:32 PM, Eric Gerlach wrote:
>>> Here's what gets me about this whole thing:
>>>
>>> - Cost to install fibre: fixed wrt GBs used
>>> - Cost to install copper to the home: fixed wrt GBs used
>>> - Cost to install all the networking equipment: fixed wrt GBs used
>>> - Cost to maintain equipment: fixed wrt GBs used
>>>
>>> All of the above are fixed wrt the GB used.  They are variable wrt Mbps
>>> offered.
>>>
>>> Quality of service has been in networking equipment for years now.
>>> So, when someone goes over their 25GB limit, don't charge them more,
>>> just throw them into a lower QoS tier.
>
> But it's not the same cost per GB.
>
> It depends upon the expected utilization rate. Both from the home, and in
> aggregate at each point of aggregation up the line.
>
> You cannot deliver fibre to the home, then assume at the first point of
> aggregation that all connections will use the full bandwidth at full rate
> 24x7, your aggregation point could neither take it all in, nor forward it
> up. At the next level up, repeat. [The same issue exists in every
> installation, your switch backbone is always less than the total possible
> aggregate of the incoming ports, let alone the switch fabric. And you can
> buy the same switch, for more $$$, with faster backbones.]
>
> So, the question becomes ... what rate do you assume?
> - and you don't dare be wrong, because the cost of fixing it, by expanding
> an installation and running another fibre to it, is prohibitive.
>
> Sort of like the situation we're in now, with copper, and increased
> (video) usage.
>
> Now you can argue they run multiple fibres, and so on and so forth - my
> only point here is that the cost is not fixed. And you're peering into the
> crystal ball, in some fashion, at some point.

My whole point is exactly that: you have fixed capacity.  So degrade
your bandwidth hogs first when you get near congestion.  Network
congestion problem solved, no need for an economic disincentive.

Cheers,

Eric