Writing on the wall for Peer to Peer?
Friday 13 September 2002 10:25:00 by
has an interesting article on the future of peer to peer file sharing systems.
The long and short of the article is that the University of Chicago has looked at the bandwidth use for various P2P applications. They have found that P2P traffic (e.g. KaZaA) can account for upto
60% of the traffic on a service providers network.
Broadband, by its very nature, will have exploded the growth of these traffic levels. We know that around Autumn 2001, BT Openworld was seeing about 25% of its traffic as P2P. There is no reason to
doubt it is similar on other ISPs in the UK.
Users will contend that they have purchased an unmetered service and can use it how they wish but
there will be a point at which on shared networks it is not possible to sustain a continued
growth in one protocol set.
Often the first casualty in any busy network is the UDP traffic (protocol used for gaming), and other low priority protocols. Of course, ISPs can grow their networks in size to ensure good
performance, and most do grow. The concern is how far can the ISPs go before deciding that they can't afford to effectively keep users on 1:1 bandwidth contention levels. At present, the vast
majority of users are still expecting and experiencing a 1:1 service but some are seeing drops to 250-300kbps at peak times.
Some solutions are already appearing, blocking P2P from part of an ISP network, low and high use tarriffs and network protocol prioritisation (i.e. making sure one protocol/type of application
doesnt swamp the network). One thing is for sure is that if P2P continues in its growth as broadband increases in popularity, ISPs will be forced to spend more money on more intelligent and faster
routing hardware. This is aside from the charges that are levied for their upstream bandwidth.