WORTH READING No likely risks that could overload the Internet

[David Farber <dave () farber net>]

________________________________________
From: Dave Burstein [daveb () dslprime com]
Sent: Wednesday, April 09, 2008 10:04 PM
To: David Farber
Cc: Brian.Randell () ncl ac uk
Subject: No likely risks that could overload the Internet

Dave

Very sorry to be writing so often and at such length, but the London Times article (and a similar one in the Guardian) 
are based on dangerous misinformation.

   The net isn't slowing down, and nearly no technical experts believe major "overload" problems likely on the 
backhaul, core, or decent local loop. Wireless is a tougher question, and small outfits are squeezed, but 90+% of U.S. 
Internet traffic is carried by large outfits who all say there is no such problem nor one likely in the foreseeable 
future. Traffic is going up fast, Moore's Law has been allowing carriers to affordably add capacity about as quickly, 
and speeds across the net are generally going up.

I'm on a crusade to focus on "evidenced-based policy.    I'd be fascinated to see any hard data that the Internet is 
slowing down or having major capacity problems. I'd also welcome any careful estimates of traffic as well as the 
capacity to handle that traffic that suggest likely problems in any well maintained network.

   Net traffic per user, as documented by Odlyzko and Cisco, has been growing  at about 35-40% the last five years, and 
that growth rate is flat and possibly down the last two years. The net has been able to handle the increase without 
price increases, much less overload, because the primary and rate limiting equipment (switches, routers, WDM, etc.) 
have simultaneously been going down at a similar 35-40%.  Moore's Law is bringing costs down and capacity up at a 
remarkable rate.

    There are problems because people will want even higher speeds than current cable and DSL (Nemertes - read the 
study not the press release). There are also some limited problems in shared local upstream cable loops, most being 
easily remedied. (DOCSIS 3.0 is at least 4-20 times as fast. Comcast announced 20M homes servable in two years.)

There are a pile of very well paid lobbyists, pr types, D.C. "policy" experts, and others essentially without facts 
approaching any reporter and crying that the Internet is facing a crisis and slowing down. A lobbying budget of 
$millions produces so much noise many honest people get confused, especially reporters who aren't technologists.  The 
top folks for the telcos in D.C. are paid literally $2M a year; for that money, you should be able to get "spokesmen" 
able to convince many that 2+2=5. They follow with a claim that telcos needs higher profits, less regulation, and 
almost everything else they want as an "incentive."

    I can think of at least 25 senior technologists, including CTOs of the world's largest networks and distinguished 
professors, who are pretty clear they don't expect major problems.

    There is a real problem for smaller ISPs in the U.K. who don't have their own backhaul, because near-monopoly 
supplier BT Wholesale is charging them an outrageous price. They won't give me hard cost figures, but the BT markup is 
at least 500% and may be as much as 1500%. (I'm estimating BT costs on the actual backhaul costs of several carriers, 
including one with three million subscribers in a comparable European country and another closer to BT's size in North 
America.) Many of the smaller ISPs like PlusNet are complaining about a squeeze, but that's a problem of weak 
competition and regulation, not a network issue. They just don't have necessary scale to do their own backhaul. BT has 
the capacity available. The likely result is that all the medium-sized ISPs could be in trouble, while those large 
enough to buy fiber will likely dominate (Carphone, Sky, Virgin cable.)

    Many of us are watching traffic closely, because we all know how video could be big, but it's simply not in the 
data for 2007 for the U.S., Japan, or most of Western Europe. You'd virtually have to have everyone in the UK turn off 
the Sky satellite and throw away all their antennae to generate enough Internet traffic that we'd see an actual 
slowdown. In the U.S., you'd need tens of millions of people to turn off their cable or satellite TV in the next few 
years to have Internet demand likely to overwhelm what can be handled.  Try running the numbers with any reasonable 
projection for video over the net.

Dave - can you or any of your readers think of any technical expert who's looked closely at this and expects major 
congestion problems in the likely future on the wireline or backbone net? I'm sure there are some, but I can't think of 
one.

Among those who have publicly and confidently said they can handle the load are  Verizon VP Tom Tauke, AT&T VP Jim 
Cicconi, and their peers at Comcast (except local loop until DOCSIS 3.0), Level 3, Qwest and and Verizon/MCI.  Also 
Cisco and Alcatel senior executives, as well as Comcast's  FCC filing on Net Neutrality.  I've heard privately the same 
from senior people in the U.K at the three largest networks. Xavier Niel at Free.fr tells me he is handling three times 
the U.S. traffic load and not having any problems.)

    Folks: I'm a reporter, not a trained engineer. I perhaps have learned enough to have an informed opinion, but 
mostly I've gone out and asked questions of the real experts as well as examined ale the data I can find. Let's not 
treat a quickly written newspaper article as evidence. Instead, let's look at real traffic numbers and how they change. 
Let's get opinions from actual experts, not lobbyists and lawyers. Let's investigate why so much misinformation is 
getting spread.

    Good policy begins with accurate information and careful analysis.

Dave Burstein

-------------------------------------------
Archives: http://www.listbox.com/member/archive/247/=now
RSS Feed: http://www.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com