[LINK] 'true' cost of bandwidth?

Scott Howard scott at doc.net.au
Sun May 3 18:58:27 EST 2009

On Sun, May 3, 2009 at 1:17 AM, Leah Manta <link at fly.to> wrote:

> If I have a 100 Mbps pipe and I put nothing over it, I'm still paying for a
> 100 Mbps pipe.  If I move 100 Mbps data across it, I'm still paying for a
> 100 Mbps pipe.

Right.  But in a commercial situation, if I've got a 100 Mbps pipe and I put
nothing over it then tomorrow I can downgrade it to a 50 Mbps pipe and pay
about half the price.

If I've got a 100 Mbps pipe and I'm sending 100 Mbps over it, then tomorrow
I'm going to have to upgrade it to 200 Mbps, which is going to double the

(Simplistic example obviously, but I'm sure you get the idea)

So whilst the volume of the data going over the link isn't directly what I'm
paying for, there is a direct correlation between the two.  If an ISPs
customers double the amount of data they are downloading, the costs for the
ISP are clearly going to go up.

So why do we pay for data caps, and data transferred, when it's really
> irrelevant to the cost of providing the capacity in the first place.  Why
> not just pay a single flat monthly fee for capacity.

The consumer ISP market has always functioned on an oversubscription model -
you sell significantly more bandwidth to your customers than you actually
buy yourself, in the knowledge that your customers - on average - never use
anywhere near the total bandwidth you're selling.  Of course, there will
always be a small number who will use excessively above average bandwidth,
and it's those users that become the issue.

In the US to date these users haven't been a huge issue.  Even if a 1.5Mbps
DSL customer runs their link at 100% 24x7, the bandwidth cost of that
customer to the ISP is less than US$5 - not enough for the ISP to care
about.  But if a cable customer runs their 50Mbps link at 100%, 24x7, the
cost to the ISP will be significant.

In effect, the US is arriving at the same position that Australia and other
countries have been in for years - it's just taken the consumer links to get
faster before it became an issue.  In Australia, a 1.5Mbps link running 100%
costs the ISP far more than they are charging.  In the US, the links needs
to be an order of magnitude faster before it becomes an issue.

There's only 3 ways to handle situations like this - ignore the problem and
lose money on those customers, only sell slower links to artifically
throttle your users, or implement some form of volume-based


More information about the Link mailing list