I’ve just been over to the BBC Internet blog to have a catchup and saw this from Ashley Highfield. In summary iPlayer has been quite a success to date, but there is still some debate about the cost to the network infrastructure.

On this point I’m glad as this debate needs to be had, both from the commercial point of view and also to inform the users of the Internet within the UK so that this is some transparency about this.

Telco2.net are also looking at this and have put out some figures based on measurements from Plus.net. I’m not going to reproduce the figures here, so I’d advise having a look over them directly, but they do illuminate one thing: the cost of carrying all this content is going up, dramatically.

This is not a surprise really, it’s been pointed out before that the cost of providing the service is not fully transparent and that the consumer is not necessarily keeping up with the cost. Plus.net’s breakdown of the data is very illuminating, especially the illustration about where the ISPs get their service from, which is essentially BT’s IPStream product. The bit I find fairly horrific is the scaling, which is done in steps of 155MBs (implying an ATM network). This has highlighted a detail in the network topology that may in fact be the weaker link.

In essence, the fact that you or I might download Gigabytes of data over our ADSL has a corresponding effect on the ISP’s connection to whichever network(s) it is peered with. If you are peered directly to the content provider, such as the BBC for example, then the amount that can be transferred between your ISP and the BBC is largely dependent on the links between them and the BBC, the cost also being negligible. So far, so good.

Now for ISPs their connectivity to you comes in via IPStream, which essentially gives them a link to the outside world and the exchanges, and for a lot of places is possibly the only option for connectivity. IPStream has  a fixed amount of bandwidth and will connect you to the backbone of BT’s network, which is being upgraded to be 21CN in anticipation of increased IP network use (!!!).

So here is the crunch, how do people pay for this ? Well largely it appears that the “Pay as You Go” model appears to be de rigueur, use more, you pay more. Easy enough. However there is the issue of scaling, looking at the figures it is fairly likely that an sustained increase would out-pace new development of the network which would impact all the users of it. This is not just the people at home, but those at work, as many businesses use the network as well. All this would appear to open the door to traffic shaping, diverting bandwidth to premium or localised services which would have the effect of cutting the Internet up into so many fiefdoms. Anyone not already on a premium connection might not get the full experience as everyone else. But then, are they now ?

One question that strikes me is how on earth are we stuck with this idea of 155MB increments ? Cable networks in Europe are looking at installing (D)WDM networks in order to cope with increased content distribution caused by iPlayer like offerings via the STB. In a lot of cases they already exist. Is this the real issue behind the scenes, lack of a decent network backbone with increased competition ? But then, do we want to repeat the experiences of NTL, telewest, Cable and Wireless, and Colt as they struggled to get their networks into the ground ?

Certainly the Cable networks will cope better by offloading these services to their VoD systems, but as yet I haven’t seen a network that can cope with every user hooked up to the VoD offering at the same time. Still, this is all services that are offered via your TV which is still a better viewing medium for most people.

There are more questions then answers, but one thing is for sure, the Internet is probably remarkably good value for the consumer at the moment, better get it while they’re still working out the bill!