Ars Technica put up an article last week about The Venice Project - a peer-to-peer Internet TV service/network/application, which is currently in Beta.  The article is a good read, and discusses how this technology could be the one that really forces the whole “Net Neutrality” aspect into the limelight for your average Joe-bob consumer.

Basically, the application of this type of service, could easily result in a typical TV watcher, pulling down about 25GB of data per month. This, when looked at from an ISP’s perspective, is potentially huge amounts of data. 

How can ISP’s handle these huge volumes of data? There’s two ways that are typically used by an ISP: Buying more network capacity (expensive) and/or throttling traffic that they see using too much capacity (cheaper).

There’s also a third option, which no major ISP is implementing on a large scale (for peer-to-peer services) - caching. 

Let me step back for a second. Firstly, head over to Wikipedia, for a quick rundown of how Bittorrent downloads and shares files.  Done that? Good.  Typically a Bittorrent client pays absolutely no attention to where another peer is - all that it cares about, is that the peer has the piece of file that it is looking for, and is capable and willing to send it. Beyond that, well, it doesn’t care. 

All (or at least most) of the popular Bittorrent clients support Cache Discovery Protocol. In essence, this allows ISP’s to set up dedicated servers, which save the content of that Torrent, and allow other users on their network to fetch it from within the ISP’s Network. Internal network capacity is typically cheaper to provide than

There’s already one commercial product by Cachelogic called “VelociX(tm)” - though this seems to only be aimed at specifically targeted content, rather than any torrent. 

Depending upon the specific network topology that an ISP uses, it may be possible to deploy relatively cheap off-the-shelf hardware to regional centres, to cache this traffic at an exchange or regional centre basis. Even for an ISP which just deployed this technology “near” their Internet Gateway, this could save significant network capacity.

The end result being that ISP’s save money (real money), and users get their files faster.  Expand this out to a project like The Venice Project, and it can result in a significant cost savings for ISP’s. I honestly can’t see a down-side to it.

Just for kicks - lets run a few numbers.
Say you spend $2m per month in Internet connectivity for your ISP, and Bittorrent traffic takes up 25% of that network capacity (Reports from some Australian ISP’s are that it can be up to 75% in some cases). Spending $200k to deploy/support/maintain several CDP servers, over 24 months (assumed lifecycle of 2 years) comes to about $8300/month.  
If that saves you even only one quarter of that total Bittorrent traffic, that comes to $125k reduction(25% of 25% of $2m) in network costs per month, or a total “cost avoidance” of over $115k per month.   OK, so you might still be paying for that network capacity which is now idle - but if it lowers the network consumption, it delays the need for network upgrades further down the line.