The National Broadband Crisis
As a follow-up to my How fast is your Internet? piece the other day, let’s have a look at the reasons behind why our consumer-grade connectivity is in the state that it’s in.
First of all, contention.
Let’s peg 8 megabit ADSL as our benchmark: I know a lot of people have less than this, but as it’s the most commonly-advertised variant of ADSL, it makes for a good illustration.
If there was no contention—that is, everybody had the full 8 megabits to themselves—you could fit the traffic from just 125 broadband users into a gigabit upstream connection. On my local exchange, there are twenty-five thousand residential and nearly two thousand non-residential premises.
Nationally, 65% of households have broadband, so let’s see what would happen if 16 250 households on my local exchange had 8 megabit connections. The numbers are simple: 16 250 multiplied by 8 = 130 000 megabits, or 130 gigabits of transit.
Okay, so that’s an awful lot of transit for one exchange (admittedly, in a fairly large town), and that doesn’t even cover the whole population. What if we introduce contention? That is, we assume that not everyone will be using their connections to their full capacity at the same time.
Given a contention ratio of 50:1 — the norm for a residential connection — the transit allocated to those sixteen thousand households drops to a mere 2.6 gigabits. If everybody was to saturate their connections, they’d get about 160 kilobits each — a far cry from the 8 megabits they were paying for, and not enough for a decent-quality video stream.
In reality, it’d actually be more than this, because each ISP would have its own transit in place (even if they all leased it from BT Openreach), but that introduces some unnecessary complexity into the numbers.
Given that industry expectations have pretty much always been that higher-bandwidth Internet services will become increasingly popular as higher-bandwidth connections are rolled out, the numbers simply don’t add up.
Let’s look at what happens if we’re a little kinder to consumers wanting to make use of their connections. How about 20:1? Well, that gives us 6.5 gigabits — or 400 kilobits per user. Considerably better, but obviously still fairly tight.
How about 5:1, then? Well, this gives each user a minimum of one and a half megabits, which is much more reasonable, but requires 26 gigabits of transit into the exchange—ten times more than they would be getting now.
In the next part, I’ll explain why feeding ten times more transit into a local exchange is so big a problem.
