Quote:
Originally Posted by Qtx
No doubt it will come in useful, no matter how/where they implement it. I really would like to know for sure if they are doing it per customers line though  |
Hi there,
Trying not to go into too much technical detail, and at the risk of opening another can of worms...
Customers' downstream traffic passes through a switch that applies traffic a ToS marking based on a pre-defined signature, source IP, port or a combination of these and other factors.
Those already with us will be familiar with the usage breakdowns we give you that show how much Usenet, FTP, Email, Streaming etc. you've done each month. It's these switches that allow us to provide you with that data. It's also these switches that apply rate limits on the lesser, 'non-Unlimited' account types.
The traffic then passes through an edge router that terminates your connection. The edge routers have a number of traffic queues (bronze, silver, gold etc.). Traffic is filed into each of these queues dependent on the ToS marking. These routers have a number of parameters for each queue including a minimum guaranteed bandwidth, a queue waiting, a maximum bandwith per user and maximum bandwidth per ‘end-point’. I wouldn't expect these values to mean anything to a customer.
We've sufficient bandwidth/capacity to allow an Unlimited user to download at line speed irrespective of the queue their traffic is in.
If a user is saturating the line using a variety of different-priority traffic then the queue weightings kick in on the edge routers and packets are dropped for the lower priority stuff.
It's this logic, and having sufficient capacity, that gives the impression of QoS on a per-user basis. In reality (and based on what I've written above) this is not strictly the case, but it's probably the explanation that best fits the actual customer experience.
Hope that makes it a bit clearer.
Best regards,