NNSquad - Network Neutrality Squad
[ NNSquad ] ISP costs [was Re: Canada goes crazy]
Vint Cerf wrote:
> resources ARE consumed when you access the Internet. The access provider
> has finite capacity to and from the network. To the extent you are
> competing with others for a shared access resource, your use prevents
> others from having unlimited data rate. So instantaneous bandwidth is
> the key resource that is "consumed" second by second.
Although I think this comes straight back to the artificial scarcity
that Bob F. keeps talking about - if you fail to provision your network
appropriately then all you have to do to make more money is claim there
is a bandwidth scarcity and jack up prices because of the "increased"
demand.
Keep in mind that what we are talking about here is a service, not a
goods market. There is nothing durable transferred to an ISP's
customers, so what are the costs an operator incurs in the course of
executing their business?
Figuratively thinking aloud, I think we can describe the following costs:
- capital costs resulting from network deployment, depreciated over
some relatively arbitrary return on investment lifetime.
- operating costs, which can be expanded as:
1. maintenance costs of the network (replacing and repairing
defective equipment)
2. support costs (employing people to assist customers)
3. administrative costs (billing, network management, etc)
4. energy costs (HVAC for all the humans and some equipment, of
course, but also energy required to keep equipment running - that is,
bits/electrons/photons moving)
Of those four operating costs I think the first three are all relatively
constant for a given network size and customer base. Energy costs,
however, could change "per bit" depending on whether network equipment
consumes very much more power while transferring more bits compared to
fewer bits - although I'm not sure if this is the case (infrastructure
equipment designers tend to be a bit lazy about power consumption and
with synchronous digital hardware then I can see that increased data
transfer might not actually make much difference to power consumption).
Then there is also the problem where there is some data shuffling around
all the time keeping various parts of the network updated.
Interesting, no? But I'm not sure what to do about it.
Russell.
[ Your list appears to omit data peering/transit issues, which
depending on the situation and particulars can involve very
significant costs. Here's an interesting paper on an aspect of
this:
"Peering vs. Transit: Why care about Transit Pricing?"
http://bit.ly/b7RD1K (Dr. Peering) ["Paging Dr. Peering ..."]
-- Lauren Weinstein
NNSquad Moderator ]