Begin forwarded message:
From: "Bob Frankston" <bob37-2@bobf.frankston.com>
Date: November 21, 2007 11:45:04 PM EST
To: <dave@farber.net>, <ip@v2.listbox.com>
Cc: "'Dave Burstein'" <daveb@dslprime.com>, "'Dewayne Hendricks'" <dewayne@dandin.com>
Subject: RE: [IP] Re: worth reading -- Internet Clogged 2010? Report doesn't say that.
The title of the report is “The Internet Singularity, Delayed: Why Limits in Internet Capacity Will Stifle Innovation on the Web” http://www.nemertes.com/internet_singularity_delayed_why_limits_internet_capacity_will_stifle_innovation_web. You can argue that the conclusions (section 9) are not that sensational but the tone throughout the report and they assume that that high capacity is the limiting factor when I argue strong it’s the lack of wire-indifferent ubiquity and their recommendations actually exacerbate that problem
As I said I didn’t want to respond to the report point by point. It presumes a broadband model with high bandwidth, AKA, video applications driving the dynamic. Even if you agree with that the presumption that more broadband is the solution it fails to look at what the real bottlenecks are in today’s distribution model of broadband and the implications of Moore’s law improvements in edge technology and architecture. Translating a demand for more connectivity into more DSL and Cable is far too narrow a claim. This is akin to the modem crises in the sense that the solution was not limiting modems but removing dial-up as the constraint.
The real limit on innovation is not a lack of high speed connectivity as much as broadband itself – DSL and Cable – which confines innovation to the billable paths owned by the carriers who set the price and configuration hurdles. The inability to presume (wireless) connectivity is a far more important limit and the proposal of more carrier-provided capacity is will continue to thwart innovation. The report also assumes wireless connectivity in the carrier 3/4G and WiMax model. No mention of pervasive Wi-Fi Connectivity. We see this with the Kindle that uses Sprint’s EVDO network thus creating a tax on every book and the user is unable to extend connectivity and forget traveling outside the US.
There is a fundamental problem with a methodology that projects from an architecture which is still repurposed telephony rather than one that is composited from local connectivity. More to the point, the Internet is about opportunity and discovering what you can do with it. So we have a broadband system explicitly designed for video distribution so it should be no surprised we use it for video distribution but it’s wrong to say that means it’s all about more video distribution. Telephony is about planning and guarantees so you have to build capacity for the applications you can anticipate and which generate revenue. But even if you accept that the future is more HDTV – that doesn’t mean more HDTV channels in broadcast mode. It just means HD bits delivered and we’re distributing prime time content over the 1% of the broadband path we get for “Internet”. A number of communities around here have three broadband paths with redundant content. That’s a huge amount of latent capacity even before we change the architecture and assuming it’s still about TV.
If you look at the analysis of wireless the report actually underestimates the potential capacity by orders of magnitude by limiting the analysis to the Telco-centric model and ignoring all those access points with megabits per access point. That cuts both ways – does that add “access” capacity or does it increase the traffic and strain the capacity as it would if you assume a broadband distribution model?
The history of the Internet has shown that demand creates capacity as we become more adept at using the infrastructure and understanding how to take more advantage of it. While I argue we’ve barely begun to take advantage of the available edge capacity (the report seems to assume that backbone capacity is not a problem) what is the “investment model”. Given the orders of magnitude uncertainty in the report it’s surprising that they are so willing to put down numbers – it’s like converting 1000Km to 621.369949495 miles. The bigger issue is where the investment is to be made and how. If we presume an investment in local infrastructure in response to perceived needs than we have a model that adapts nicely to growth – those who have a problem invest in capacity. If we accept a Moore’s law model for adding capacity the cost is already very low and will become too trivial too notice –it’s already far less than just about any other municipal infrastructure we have. Just look at the cost of connecting your house to the electric grid -- I’m upgrading my service now and it’s a lot more expensive than a fiber connection would be using current technologies. It’s an 1/8th trench vs. a huge pipe with thick wires and all sorts of other challenges.
But the report says “The only hope to close the gap permanently is that technology will provide access solutions that are less capital-investment intensive. Improvements in DSL technology, the rapid build-out of optical access and the emergence of wireless alternatives the last mile will help, but ultimately, access investment by the carriers will be required in order to address the big crunch from a supply side. “
That is not just wrong but dangerously wrong – it’s like investing in the foxes to increase the number of hens. And what is “supply side” for the Internet?
They do have disclaimer in explaining sensitivity analysis and given that their model is flawed the results should be viewed modulo that model. Digital systems make modeling especially difficult – the report says that small changes in utilization result in major increases in demand – but it cuts both ways. But, as I said, Moore’s law is about supply not demand.
The report itself is long and full of qualifiers and I presume the authors are open to counter arguments. The danger, however, is these kind of reports are not benign – they feed lobbyists’ need for credibility. All that they need is a conclusion that says the poor carriers need more money and incentives to save the Internet. That can do real harm which is why I feel obliged to respond.
From: David Farber [mailto:dave@farber.net]
Sent: Wednesday, November 21, 2007 22:05
To: ip@v2.listbox.com
Subject: [IP] Re: worth reading -- Internet Clogged 2010? Report doesn't say that.
Sent: Wednesday, November 21, 2007 22:05
To: ip@v2.listbox.com
Subject: [IP] Re: worth reading -- Internet Clogged 2010? Report doesn't say that.
Begin forwarded message:
Date: November 21, 2007 9:57:44 PM EST
Subject: Internet Clogged 2010? Report doesn't say that.
Bob, Dave, Dewayne
Re: "Oy again? Wasn’t the phone system supposed to collapse due to modems? The report is too full of nonsense to rebut in detail."
Lots of things to disagree with in Johna Till Johnson's report, but most of the news stories and comments on the list clearly didn't reflect the report. For example, in the summary she says
'It’s important to stress that failing to make that investment will not cause the Internet to collapse. " and "core fiber and switching/routing resources will scale nicely to support virtually any conceivable user demand" which doesn't sound like "collapse" to me. Similarly, "The impact of inadequate access infrastructure is likely to be relatively mild," in the conclusion.
Nemertes' primary conclusion is sound. People will probably want more local
access - faster DSL and cable. This is almost certainly true - many people today want more than the maximum 1 meg up, 3-6 meg down that the telcos offer for DSL. Cable upstream is typically even slower. That's a small fraction of the bandwidth needed to send grandma the DV video from a $400 camcorder. As more people move to HD and multiple TVs, as well as other bandwidth hungry apps, many will want higher speeds.
The pr people stretched this. possibly out of ignorance, including a headline "User Demand for the Internet Could Outpace Network Capacity by 2010." "The Internet" is the network of networks, the part that "will scale nicely." The pr (partially paid by AT&T) added a uninformed comment from Larry Irving "a critical issue facing the Internet ...potentially face Internet gridlock that could wreak havoc on Internet services.” That's directly opposite the actual conclusions of the report, that the Internet does fine, and the local loop will have speeds higher than today's (except possibly with obsolete shared cable modems.) The press release, but not the report, had the Howler "it may take more than one attempt to confirm an online purchase". Confirming an online purchase requires a small number of bytes and is not very sensitive to latency or even a significant slowdown.
Some careless reporters then took it much further, including transforming "could" but unlikely to "will," as if it were almost inevitable. If his paper were not in the midst of a round of layoffs, I'd single out one particularly bad story.
Author Johna Till Johnson and Sponsor Larry Irving are acquaintances who have done work I respect. Johna should have been far more careful about material attributed to her organization, and mistaken comments by her partners and reporters. We also have honest disagreements. I believe some of her numbers, like typical Internet usage today, are much too high. Because the carriers hide the data, neither of us can be sure who has it right. I also believe she's reading far too much into a simple model, clever as it may be.
Former government officials Larry Irving and Bruce Mehlman are partners in the "Internet Innovation Alliance" , which bills itself as a non-profit with broad support. I believe in fact most or almost all their money comes from the carriers and their suppliers. Mehlman is a well-paid registered lobbyist for AT&T, and they consistently advocate positions useful to the Bells. IIA may have found a loophole that allows them to skirt the law on registering as lobbyists, but it's disingenuous to suggest they are independent.
All of which makes me wonder why I, Karl of DSL Reports, and others have spent hours trying to get the correct facts out. We should be able to bypass foolish mistakes like this. But folks like Senators and FCC Commissioners far too often believe what they read in the papers. That's one reason even honest policymakers (Mike Powell) make profoundly wrong decisions. No one has calculated how much the telcos are spending "creating a climate of opinion." If the number came out, it would be shocking. My best guess is upwards for $700M is spent each year to influence the FCC and policy, but I can't come close to proving that.
Dave Burstein
Perhaps the absurdity is evident in their statement “This resulted in the first-ever study that assessed both infrastructure investment and current/projected traffic patterns independently, and compared the two. It is also the first study to apply Moore’s Law (or something very like it) to the pace of application innovation on the “Net” and validate that it appears to conform to the available data so far.” They are willing to apply Moore’s law to increasing demand but not to the ability to take advantage of existing technology. This is completely backwards – the point of the end-to-end constraint is that we take advantage of opportunities – the applications adapt to what is available and it’s not just about more video from a central source – just the opposite! The highest value applications put the least demand on the infrastructure! (OK, TV has high value but…)
They do say that the backbone will scale but the problem is in the access network. Their solution, no surprise, is more of that Olde Tyme Broadband. Of course the solution is more money for “broadband” even as the industry works at a furious pace to prevent a repeat of the overcapacity of the “fiber bubble”. It’s as if we need to pay the carriers from continuing to keep their capacity unavailable and wasting it in using their last mile networks in broadcast TV mode.
"Want a seamless end-to-end network?" was the tagline for the ad that greeting me going to the site. Of course that's a womb-to-tomb model. It gives a sense that this is part of the IPSphere mentality warning us that we need protection from the wrath of the network owners if we deny them the money they demand to fund their networks.
-----Original Message-----
From: David Farber [mailto:dfarber@cs.cmu.edu]
Sent: Wednesday, November 21, 2007 18:22
To: ip@v2.listbox.com
Subject: [IP] Internet could clog networks by 2010, study says
Begin forwarded message:
From: dewayne@warpspeed.com (Dewayne Hendricks)
Date: November 21, 2007 10:37:44 AM EST
To: Dewayne-Net Technology List <xyzzy@warpspeed.com>
Subject: [Dewayne-Net] Internet could clog networks by 2010, study says
Internet could clog networks by 2010, study says
Telephony Online
By Sarah Reedy
User demand for the Internet could outpace network capacity by 2010,
according to a study released today by Nemertes Research. The study
found that corporate and consumer Internet usage could surpass the
Internet access infrastructure, specifically in North America, but
also worldwide, within the next three to five years.
< http://telephonyonline.com/home/news/internet_network_capacity_111907/ >
-------------------------------------------
Archives: http://www.listbox.com
They do say that the backbone will scale but the problem is in the access network. Their solution, no surprise, is more of that Olde Tyme Broadband. Of course the solution is more money for “broadband” even as the industry works at a furious pace to prevent a repeat of the overcapacity of the “fiber bubble”. It’s as if we need to pay the carriers from continuing to keep their capacity unavailable and wasting it in using their last mile networks in broadcast TV mode.
"Want a seamless end-to-end network?" was the tagline for the ad that greeting me going to the site. Of course that's a womb-to-tomb model. It gives a sense that this is part of the IPSphere mentality warning us that we need protection from the wrath of the network owners if we deny them the money they demand to fund their networks.
-----Original Message-----
From: David Farber [mailto:dfarber@cs.cmu.edu]
Sent: Wednesday, November 21, 2007 18:22
To: ip@v2.listbox.com
Subject: [IP] Internet could clog networks by 2010, study says
Begin forwarded message:
From: dewayne@warpspeed.com (Dewayne Hendricks)
Date: November 21, 2007 10:37:44 AM EST
To: Dewayne-Net Technology List <xyzzy@warpspeed.com>
Subject: [Dewayne-Net] Internet could clog networks by 2010, study says
Internet could clog networks by 2010, study says
Telephony Online
By Sarah Reedy
User demand for the Internet could outpace network capacity by 2010,
according to a study released today by Nemertes Research. The study
found that corporate and consumer Internet usage could surpass the
Internet access infrastructure, specifically in North America, but
also worldwide, within the next three to five years.
< http://telephonyonline.com/home/news/internet_network_capacity_111907/ >
-------------------------------------------
Archives: http://www.listbox.com
