North American p2p went from 370 petabytes in 2006 to only 416 petabytes in 2007 according to Cisco’s figures. Since U.S. users increased 16% in the same period, that’s a drop in p2p per user and a significant drop in p2p as a percentage of all traffic. There’s a major margin of error in these figures, so I’m calling it “flat.” That’s very different from pre 2007 experience, when p2p grew rapidly. It severely contradicts what many in Washington D.C. are saying, but who would believe a lobbyist? In previous reports, I have mentioned two sources that indicated p2p had dropped to about a quarter of traffic. P2p isn’t down everywhere—see below for a contrary datapoint and opinion. Cisco projects that p2p in the next five years will grow at less than half the rate of Internet traffic. P2p outside of North America has and will grow faster, but also more slowly than overall traffic growth. I can confirm from European sources that they have not seen the p2p traffic slowdown seen in at least some North American carriers.

Cisco’s 2007 estimate for U.S. p2p is a little less than half their estimate for total consumer traffic, dropping to about 40% in 2008. Europe, they also believe, was a little less than half in 2007, but the p2p percentage they expect to increase in 2008. Asia they place at 57% in 2007, dropping to about half in a few years. Denny Strigl of Verizon just said “about 60 percent of Internet traffic is conducted on a peer-to-peer basis,” but Denny’s a wireless guy and not yet a trusted source on wired connections.

A reader at a world-class carrier is having a very different experience on his network, with rapidly growing p2p. Some of the difference is that encrypted traffic is up significantly, and he believes the vast bulk is p2p and counts them that way. I’m touching base with Cisco to look deeper into their numbers, which overall are the most carefully produced publicly available figures. There are some gems in the tables here and here.

Overall, traffic growth seems flat to down. Nikos Theodosopoulos finds “we are seeing some signs of moderation in growth rates.” Cisco’s projection for Internet traffic growth is 43% overall, which is 33-36% growth per user, the figure you need for costs and policy. (The 46% total IP bandwidth in the headline is the right number to use if you are, for instance, forecasting total demand for IP service provider routers. It includes the IPTV video build-out at AT&T, cable VOD, etc. My 43% figure is a quick adjustment of Cisco’s data to look at the Internet growth rate.) 43% overall would be 33-36% growth if it were all on the wireline side, but both Cisco and I expect wireless data demand to grow much faster. The per user traffic growth on DSL and cable networks, working from Cisco’s figures, should be 25-32%

25%-32% growth sounds enormous, but actually is right on target and possibly below the trend of the last seven years. Cisco sees growth “continuing to decline in subsequent years,” after a spike in 2008. (Except for small pockets—Africa, cable upstream until DOCSIS 3.0 comes along, subscale providers—Internet speeds are increasing and congestion decreasing unless the carrier in question is severely cutting capital expenditures.) Something like 98 out of 100 engineers who follow the net are confident traffic growth in the future will be handled as efficiently as it has been for the last 6 years. The widely publicized fear of Internet slowdowns is a result of some of the most skilled lobbying in the world. Larry Irving, key to that effort, is proving to be the most effective lobbyist in D.C. As a Democrat and Obama supporter, he will be able to name his price if Obama wins. He should easily pull the $2M+ people like Tauke and Cicconi report publicly, although probably not the $7M Bill Barr, Verizon’s Mr. D.C. behind the scenes guy makes. They (and most of the other telco advocates) are Republicans, so the companies are offering top dollar for available Democrats. Larry is proving to be very, very good at spreading the message.

I’ll welcome ideas, on or off the record, about the trend in traffic.