IPTV/VoD: Cutting off the air supply
Just how juicy is Joost?
Let's look at some basic ISP economics
If you want to provide DSL services in the UK, you can either resell BT Wholesale's network, or put your own equipment in one of its 6,000 or so exchanges (called local loop unbundling, or LLU). The point to point copper wire running between someone' house and the exchange, via the green street cabinets (called the sub-loop) is always maintained by BT Wholesale and has to be rented.
The original incarnation of wholesale DSL products in the UK came in three flavours – Datastream, IPStream and Videostream. The latter is now defunct, and IPStream is by far the most common of all. Datastream was designed so that the ISP buys capacity to a specific exchange, rather than aggregated bandwidth on a BT Central pipe. Bandwidth works out from £57 to £91 per Mbps but means that ISPs need heavy density on each exchange to make it work, as a product or you end up with stranded capacity in certain locations.
IPStream is the main product most consumers are using for their broadband service. Adding a new customer to your network means a £40 activation charge or an £11 migration charge if they already have a DSL service and are changing provider. If you manage to persuade 10,000 people to switch to you, you have a £100,000 bill to pay. God forbid they are new customers from your fantastic new fluffy marketing campaign you just blew the year's budget on, as you'll be writing a cheque for £400,000.
For each of your subscribers, you have a monthly rental to BT of £8.40 (standard) or £12.40 (premium) for a connection of up to 8Mbps dependent on line characteristics. This is the charge for connecting from the end user's location back to the local exchange. The service is not contended per se, but rather has a complicated sharing algorithm that means under busy conditions the maximum throughput could be 400k.
"Premium" comes from the old 50:1 20:1 contention ratios used by BT's Home/Office products and supposedly means better performance (e.g. 800k vs. 400k upload, "preferential" treatment at times of low capacity). So far with our 10,000 subscribers, we are clocking up a minimum monthly rental bill of £84,000 or £124,000.
Now the fun part – backhauling data from the local exchange to an ISP's own network, where it can be routed onto the public internet. Unless you have your own national network, you'll be using BT again. For this, you need a BT Central pipe and some international transit. A Central pipe costs around £31,000 per month for each 155Mbps capacity, or £200 per Mbit.
No, you read that correctly. £200 per Mbit capacity. No, seriously.
You are allegedly supposed to be able to fit 8,000 subscribers onto that (with their individual share of the contented connection of around 30K – no really), but most ISPs generally only manage 4,000 to 5,000. With our 10,000 users, we need two pipes, so our bill is around £62,000 a month.
Trunk Internet transit, thankfully, is well priced and well served. Charges for 10Mbps are around £50 at worst case, but with a little negotiating skill and big buying (multiple Gbps) it's entirely possible to get the price down below £9. Using those figures, a subscriber watching a 400k stream 24 hours a day costs an ISP around £88.40 per month + VAT. If a subscriber was using their 8Mbps connection at full speed 24 hours a day, they could in theory potentially cost £1,600.
BT's 21CN program promises to change the old legacy ATM infrastructure to a unified IMS architecture that bases all the services BT offers on IP. The bandwidth pricing looks like it's going to be around £75 per Mbit so ISPs who want to get involved need to build their own national networks to backhaul connections from regional data centres (Points of Prescence, or PoPs). A £125 per Mbit reduction (62 per cent or so) is a step in the right direction but the catch is the estimated £5m needed to build your own national network.
But, of course, you could just ignore all this and get your own little LLU train rolling. You have a lot of freedom then, even to do fun things like TV that BT couldn't allow you to. Only one small snag. You need the GDP of a small African nation to do anything remotely sexy. You can unbundle an urban exchange for £15,000 (hardware and BT co-location fees) and backhaul it with a 1 Gig connection for £30-60,000. Then you need to do the maths and geography, which tell you to do that for the top 1,000 exchanges, you're going to need something in the range of £80m. If you give each customer a dedicated 2Mbps line, you're only going to fit 500,000 people on your network, which isn't exactly your investors' or CFO's idea of value for money.
Sponsored: What next after Netezza?