Assume those were packet loss numbers
Would anyone still be saying that 2-3% packet loss is acceptable?
By contrast, your average Internet backbone provider runs their network at less than 0.001% contention. Our acceptable level of packet loss on IP networks is exactly 0.0% -- anything more than that and interactive applications start becoming practically unusable.
Don't get me wrong, I'm less than enthused as the quality of the data that Bell has published, but there's a lot of bullshit coming from all sides. I especially like the crap being spewed about upgrading the DSLAMs to GigE access as being the cure-all. The numbers published by Bell indicate congestion is a path issue, not a point issue. If you upgrade all of your DSLAMs that data still needs to go somewhere.. that means upgrading the aggregation network, which in turn pushes upgrades to the core. This isn't a problem that you fix by taking dad's credit card to Radio Shack to buy a few parts.
But the question is, "With this proof now in the clear, will anything actually be done about this?" I wonder
LOL nothing, it's Canada... we bend over and take it.
|reply to espaeth |
Packet loss has increased throttling if you look at the provided numbers.
Now about congestion figures provided by Bell, you need to read the small text. A link is congested when it reaches 60-90% capacity (not 100% capacity) at least 0.30% of its uptime. That still means a link could be uncongested for 99.7% of the time and be deemed "congested".
Also, traffic just moved from the throttled service to unthrottled services, but with higher packet loss, following the throttling of the service. In other words, "throttling" does not reduce traffic unless you throttle the whole connection. When you're there though, it's just that you reduce the speed of all the users because you can't provide the marketed speed.
Another solution is to provide bandwidth limits (caps) that are different based on the time of the day. The cell phone industry does it, maybe that's what ISPs should do.
Anyway, the bottom line is that throttling is a stupid idea, moreso when you decide to throttle the concurrence to make sure they can't compete with you based on that point.