Over the past week, some pseudo-engineers have been sending me “fan-mail” saying that\u00a0telco aggregated\u00a0connections can’t ever be blocking because the ISPs pay for a certain peak\u00a0rate bandwidth and should be able to use this however they want. These folks are forgetting a number of aspects of internet traffic, perhaps\u00a0having been sheltered from having to actually operate a real internet access network.<\/p>\n
Gather round. This is what we call a teaching moment.<\/p>\n
The most important\u00a0lesson is\u00a0basic network engineering economics:\u00a0services are kept affordable by sharing the same network facilities among mutiple customers. The level of sharing is determined by making usage assumptions in order to avoid congestion. To the extent things change from these assumptions, new engineering and network reconfiguration needs to take place.\u00a0[Note: when you are sold a circuit capable of a specified data transfer rate, there are assumptions that you will not use that circuit around the clock at that speed.]<\/p>\n
The\u00a0wholesale price\u00a0for aggregate ISP customers connections\u00a0is kept low by combining the access network with the traffic of other internet access customers. Since the ISPs aren’t specifying the peak traffic requirements of each customer or each switching centre, it is quite possible for one customer to use a disproportionate share of the capacity, again driving a need for re-engineering of the internal network.<\/p>\n
This brings us to the current debate, which has nothing to do with retail usage based pricing.<\/p>\n
When the price of the wholesale aggregated internet service was developed, there\u00a0was some kind of assumption made about the average load being generated per end user. This was used in the production of the tariff for Gateway Access Service (GAS)\u00a0in the order of $20 per month, which covers the carriage of the end user traffic to the network interconnection point of the choosing of the ISP.<\/p>\n
The size of the pipe between the telco and ISP varies with the amount of traffic, but this can increase in two ways: by adding more customers, or by each customer using their service more. The phone company gets fully compensated if the increase comes from more customers who have a traffic profile that corresponds to the engineering assumptions; the problem arises when the\u00a0average traffic has different characteristics from the engineering assumptions.<\/p>\n
This is why the CRTC needs to revisit its Usage Based Billing decision and set a threshold based on the loads presumed in the engineering for GAS, with the tariff incorporating appropriate adjustments for aggregate traffic loads that exceed these assumptions.\u00a0<\/p>\n
I note that some ISPs had requested an ability to gather their traffic on a more localized basis and I wonder whether this would have have pre-empted the current wholesale disputes.<\/p>\n","protected":false},"excerpt":{"rendered":"
Over the past week, some pseudo-engineers have been sending me “fan-mail” saying that\u00a0telco aggregated\u00a0connections can’t ever be blocking because the ISPs pay for a certain peak\u00a0rate bandwidth and should be able to use this however they want. These folks are forgetting a number of aspects of internet traffic, perhaps\u00a0having been sheltered from having to actually …<\/p>\n