[pyre]

  > Here in a suburb of Toronto I've had top-notch,
  > leading-the-bandwidth (15Mbps around the clock) high
  > speed for a low cost for almost a decade.

Toronto is the "New York City" of Canada. Do you look to New York City to be an exact replication of how the entire US looks, works, and functions? This reeks of "Windows never gave me a BSOD therefore all of the people that say they've had a BSOD are lying"-type of argument. Please leave your anecdotes at the door unless you promise to see them for what they really are... anecdotes.

When I was living in Toronto the quality of internet service (at least with Bell) depended very much on the area. In some areas the lines/infrastructure was really crappy and rather than fix it Bell would just blame the end-user (you need to reboot your computer, that's why the bandwidth is so slow) or acts of god (I know that you've been complaining about the bandwidth for months, but it was really caused by that large storm last night. It should be fine now.).

  > Bell is far from alone in providing bulk bandwidth in Canada.

Really? I was under the impression that Bell was just selling access to their last-mile infrastructure and companies like TekSavvy were providing their own peering arrangements for getting the customers to the internet. Would you care to explain to me what the real situation is?

                +------------------------------+
                |         THE INTERNET         |
                +------------------------------+
                        ||         ||
                        ||         ||    <-- Peering arrangements
                        ||         ||
                     +--------+  +----+
                     |TekSavvy|  |Bell|  <-- ISPs
                     +--------+  +----+
                        \\       //
                         \\     //
                          \\   //
                  _        ______      
      Bell-owned |        /  ||  \
  infrastructure |       /   ||   \    <-- last-mile infrastructure
                 |_     /    ||    \
                      +--+  +--+  +--+
                      |  |  |  |  |  |  Residences/Customers
                      +--+  +--+  +--+

It's my understanding that the infrastructure looks like the above diagram. In this case, Bell is not selling bandwidth to the internet to 3rd-party ISPs, they are just selling access to their (government-subsidized) last-mile lines.

[Swannie]

Lets clarify a few things.

The ruling last year: http://www.crtc.gc.ca/eng/archive/2010/2010-802.htm

The result of consultation on the ruling, and dates for its application: http://crtc.gc.ca/eng/archive/2011/2011-44.htm

This affects the "Gateway Access Services (GAS)" customers. Wholesale Bell customers. (Product details here http://www.wholesale.bell.ca/pdfs/GASDSL.pdf)

And "Third Party Internet Access (TPIA)" customers. Rogers & Shaw Wholesale customers.

The GAS is not local loop unbundling. That is "(ADSL) Access Service" (http://www.bce.ca/en/aboutbce/regulatoryinformation/tarrifs/...)

There are essentially three options for an ISP. Diagrammed below.

PoP = Point of Presence. This is the handover from the access network to the provider network. The equipment found here is often called a BRAS.

    ------------The Internet------------
     |          |          |          |
    ----------Peering Location----------
     |          |          |          |    <-.
    ISP1       ISP2       TekS       BELL     \
     |          |             \    /      <---- Private network.
    PoP1       PoP2            PoP3
     |          |#              |**
  Priv.Fiber LeasedLine        Bell    
     |          |               |**   <---- "Backhaul" "Transmission Network"
  Priv.DSLAM Priv.DSLAM    -Bell DSLAM-
     |~         |~         |          |*  <-- "Local Loop" "Last Mile"
    CPE        CPE      TekSCPE     BellCPE

  * Bell's retail capacity based pricing.
  ** Bell's wholesale capacity based pricing.
  ~ ADSL Access Service
  # Bell does actually lease lambdas on its fiber.

Source: http://en.wikipedia.org/wiki/File:XDSL_Connectivity_Diagram_...

[pyre]

What is the section between the DSLAM and PoP? Why does Bell need to charge usage-based on that section seeing as the peering itself is not the issue? Some more explanation of that part of the diagram would be much appreciated. :)

[Swannie]

Umm... that's the bit between your local exchange (Central Office) and the nearest Bell exchange (Central Office) to where the ISP actually hosts its equipment. (In the case of the GAS, that equipment is routers that can talk PPP/L2TP, or something close, as I can't find real technical details on GAS). OK, so it is more complicated than this as there are BRAS - broadband remote access/aggregation servers - along the way that collect the DSLAM traffic together into ATM links, and the split out aggregated traffic for the ISPs. (Basically the internal Bell ATM packet network).

This is Bell's transit network. Whilst technically the UBB is on the individuals and not what is transited, it is essentially what is being billed for.

Bell is claiming that it's links on the transit network are being crippled by heavy users. This may actually be true during busy hour. But in reality this shouldn't happen, as Bell can increase the bandwidth of most these relatively cheaply, as most are fiber. There are probably a few 100 that aren't fibered, causing the biggest problem.

But OK. The more I read about this, the clearer it becomes that the problem doesn't really exist at the moment. This is Bell's way of sabotaging the GAS product that it has been forced to sell by the regulator.

I hope this is a move that will force the independents to group together to purchase their own DSLAMS and transit. And then Bell faces much stiffer competition. In the UK, BT is facing real and consistent competition, thanks to our regulator getting tough in the last few years. BT now even have to allow competitors to lease it's ducting to roll their own fiber.

EDIT: I found this: http://www.vaxination.ca/crtc/ADSL_intro.pdf

Which suggests that the only bit that could conceivably have contention, which the wholesale customer does not control, is the connection between the DSLAM and the BAS. From the BAS's to their handoff PoP/CO, they have to pay for these "AHSSPI/AHSSL (Aggregated High-Speed Service Provider Interface/Link)" in 1gbps increments.

[jacquesm]

As far as I know that's an accurate representation of the situation.

[Swannie]

It's not.

[ergo98]

My anecdote was...anecdotal. Nonetheless Canada's broadband deployment rates are in the top quarter of the OECD, and it regularly ranks very well in bandwidth in the same comparisons. Empirically Canada isn't doing so terrible.

Really? I was under the impression that Bell was just selling access to their last-mile infrastructure

Your understanding is wrong. If TekSavvy were just terminating the loop, Bell wouldn't even have the technical ability to do anything to restrict or harm them.

The situation here is that Teksavvy has access to bulk wholesale Nexxia (Bell's backbone) rates that were forced upon Bell by the CRTC. Again, this was an agreement that was established when Bell was overwhelmingly dominant in certain areas of Canada. If those wholesale rates are "fair", however, then clearly Teksavvy can just arrange with another provider to handle that service. The reason they can't, of course, is that the rates they were paying were essentially monopoly subsidized.

[pyre]

Is that really the reason why? I find it also possible that they haven't been able to setup the infrastructure to peer with other providers yet, as it seems they were only given 90 days before this went into affect. I don't know what sort of timeline is reasonable for them to be able to hook-up with other providers as I've never worked on that sort of infrastructure before.

[ergo98]

Yeah, that's the reason why. Just to be clear, Bell has been making their intentions clear for a long, long time. They were packet shaping "reseller" connectivity as long as two years ago.

There ARE DSL providers that locate hardware in Bell switching centers and don't use the Nexxia network. Primus is an example. Funny thing about Primus is that they, too, are implementing UBB, but it isn't because of Bell: It's because they don't want the ultra high throughput users that are being punted from TekSavvy from all migrating to their network.