The internet is a shared resource. Different access providers begin mixing traffic at different places, but sooner or later, my internet gets mixed into yours.

The Canadian Association of Internet Providers (CAIP) application to the Canadian Radio-television and
Telecommunications Commission (CRTC) seems to acknowledge this shared nature with its reference (at paragraph 50 of its application) to the description of the Gateway Access Service its members resell, a description complete with a graphic of a cloud—a sign that the resource is being shared.

Cloud, network, virtual channel. All of these are distinct from terms like private, dedicated or other ways to denote an unblocking and congestion-free connection. A private circuit is expensive but can be more susceptible to network failures; shared, switched or virtual circuits are more affordable and can often allow for routing around blockages.

In CAIP’s application to the CRTC, I noticed that it couldn’t resist the common fallacy of misquoting from the Report of the Telecom Policy Review Panel. (I pointed out this common problem in a blog posting more than a year ago.)

The CAIP application (paragraph 91) quoted the report as recommending that the Telecom Act should be amended:

...to confirm the right of Canadian consumers to access publicly available Internet applications and content of their choice by means of all public telecommunications networks providing access to the Internet.

But CAIP neglected to complete that passage, which was in fact one of the formal recommendations from the Telecommunications Policy Review Panel (TPRP). Recommendation 6-5 continues with:

This amendment should
(a) authorize the CRTC to administer and enforce these consumer access rights,
(b) take into account any reasonable technical constraints and efficiency considerations related to providing such access, and
(c) be subject to legal constraints on such access, such as those established in criminal, copyright and broadcasting laws.

Part (b) seems to be designed to take into account the technical realities of the internet as a shared network resource—even for the Gateway Access Service used by CAIP members to resell Bell internet.

Different applications place different demands on network capacity. Some have no tolerance for latency (delay) in delivery of messages. Two-way interactive applications like voice and gaming are examples of applications that need low latency. Others, like streaming video or music, use buffering to self-manage a certain amount of latency: the application can stockpile the next few seconds of media content in order to absorb the effects of some of the latency in the delivery from the source. For other applications, such as file transfer and email, latency can cause frustration but not loss of functionality because the data transfer requirements are not real-time in nature.

If customers want to be able to make VoIP phone calls, play network games, watch streaming video, listen to internet radio and capacity is running short, what do you do? Do we think the balance of convenience should put these applications at risk in order to shave some time off bulk file transfers?

Treating all bits the same means that real-time applications will become non-functional in times of peak traffic.

What kinds of technical constraints are reasonable in managing shared resources?