According to Juniper Networks, YouTube already generates traffic equal to the entire internet-load in 2000.
Indeed, the widely held assumption is that the explosive growth of video across the internet will quickly exhaust excess capacity and push bandwidth prices up. But so far, this has yet to be proven.
Not everyone believes that the widely reported “fibre glut” of the late 1990s and early 2000s will be exhausted by video, prompting a spike in the price of retail and wholesale bandwidth.
“In long haul, there is still plenty of fibre,” says Andrew Odlyzko, director of the Digital Technology Center at the University of Minnesota. “If you look at the total internet traffic in the US, it could be squeezed down one or at most two fibre strands. And on most routes you have hundreds of strands.”
Insight Research did a study back in 2001 of fibre utilisation among 13 major long-haul carriers. Data was culled from fibre pairs in 24 major cities. Only 7-8% of the total capacity was used, and of that, only 3-4% was actually lit, says Robert Rosenberg, president of Insight Research.
“It was really a small percentage of the capacity in the ground,” Rosenberg says. “You said to yourself, ‘Gosh, this thing is never going to go away.’”
Factor into that the slowing growth of internet traffic. Even though the rate of video growth has been increasing — Level 3 Networks says 50-60% of the traffic across its IP backbone is video, compared with 5-10% five years ago — the overall growth of traffic on the internet has slowed to 50% anually from 100% or more in the heady days of the tech boom.
“Back in those days, everybody was putting in as much as they could and it made sense,” says Clif Holliday, an analyst with Information Gatekeepers (IGI). “There’s probably an awful lot of excess fibre in the ground.”
According to IGI, internet traffic is expected to grow 45-50% per year until 2010. The major feeders of traffic on the internet backbone will be high-speed DSL and cable modem broadband access lines, international transactions and fiber-to-the-premises (FTTP) lines.
File sharing is a major component of high-speed traffic, and the largest segment of file sharing is video, according to IGI. But video file sharing will be dwarfed by FTTP traffic in the form of internet protocol television (IPTV), especially high-definition IPTV (HDTV), Holliday says.
“Using digital delivery of HDTV is a tremendous bandwidth hog,” Holliday says. “It will overshadow everything else if the telcos do what they say they are going to do, and if they are successful at it.”
Bandwidth pricing could increase, Holliday says, because carriers will need to purchase equipment to support IPTV across specific fibre routes with a shortage of capacity.“Somebody could say, ‘we’ve got a lot more fibre than we need’ — that’s probably going to be true maybe forever — but that doesn’t mean that we don’t need to have more,” Holliday says. “It doesn’t make any difference how much fibre you’ve got in total in the United States. If you need to get from A to B, the only thing that makes any difference is how much you’ve got from A to B. There will be routes where you’ll see major additions.”
Global Crossing recently increased wholesale and “selective” retail bandwidth pricing by 5-10% due to “supply and demand equilibrium”, says Anthony Christie, director of marketing for the carrier.
Christie says this equilibrium is a confluence of several factors, not just the growth of streaming video on the internet. The virtualisation and web-ification of enterprise applications, industry consolidation and inexpensive broadband access — which enables sharing of video files — are all playing a role in the stabilisation and increase of bandwidth pricing, he says.
As is the case with Level 3, Global Crossing finds that 50-60% of the traffic from its top 10 IP transit customers is “video driven”, Christie says.
Jeff Tench, senior vice president of offer management for Level 3 Content Markets, also sees prices for that bandwidth — be it transport bandwidth, IP transit or content distribution — stabilising due to higher demand from the end-user as well as from carriers for edge applications.
That higher demand has helped absorb the capacity glut, Tench says, prompting a round of reinvestment in the network. Level 3 has been investing in its network for the past three years at both the wavelength division multiplexing and IP layers, he says.
Level 3 has also been one of the more active carriers on the consolidation front, acquiring seven companies over the past two years.
Demand and upgrade costs to light unused fibre will help stabilise bandwidth prices, says Eric Schoonover, senior analyst at research firm TeleGeography. He said it would be a “stretch” to expect prices to increase based on the growth of video.
Video is driving the user’s experience, which is prompting providers to buy more IP transit bandwidth to handle the increase in traffic on their long-haul networks, Schoonover says. But this will not eat into fibre glut, he says.
“There’s plenty of fibre in the ground for years to come,” he says.
IGI’s Rosenberg disagrees, saying carriers will need to replace embedded and unused glass with newer-generation strands.
“This is what the industry has waited for since the 1960s,” he says. “Once you get a mass market for video telephony, that application is going to continue to suck up capacity because whatever you gave them last week is not sufficient: I want better definition, I want better sound, I want better quality in the image.
“It will mean that the industry will begin another round of investment with a new generation of glass and a new generation of optics to be able to meet that capacity demand,” he says. “Current unutilised fibre will be obsoleted by changes in technology over time.”