Print 50 comment(s) - last by sphyder.. on Feb 21 at 2:19 PM

Cable giant defends Data Discrimination policy with 80 pages of chaff

The Internet was set alight last October, when news broke out that Comcast actively interfered with P2P traffic. Nasty words flew back and forth. Lawsuits were filed. Even the FCC joined the party, kicking off its own investigation early last January.

Fast forward to last Tuesday, when Comcast answered the FCC’s inquiries with an 80 page reply (PDF) that explained the company’s rationale behind its definition of “reasonable network management:”

Notwithstanding the enormous capacity and flexibility of the cable infrastructure, there are (and always will be) some throughput limitations. Thus, the question is not whether all customers will be able to use shared bandwidth indiscriminately… but, rather, how to optimize every customer’s online experience and ability to … use all Internet applications and services.”

Fair enough. Other parts of the document point out that Comcast is more concerned with making sure that their average users can surf the web proper, as opposed to placating the desires of a “vocal minority of users who make the most noise.”

Right there, however, Comcast’s logic falls apart. A quarter of the way through the document – Page 14, according to Acrobat – Comcast defers to the testimony of Congresswoman Mary Bono Mack, who says that “P2P applications are designed to consume as much bandwidth as is available, thus more capacity only results in more consumption.”

So that much is true – but just as true are the numerous other protocols and applications that are also bandwidth intensive. Almost every other non-streaming application works because of similar goals: using as much bandwidth as necessary to obtain optimal results. In some cases, bandwidth requirements are minimal, like in online chat applications. In other cases, however, saturating a user’s limited pipe simply isn’t fast enough, as you simply cannot escape the fact that downloading a file or streaming a high quality video requires a comparatively large amount of data. There are many, many bandwidth hogs out there that are just as good, if not better, at “clogging the tubes” than BitTorrent and its ilk.

Look at Windows Update: every time Microsoft drops a big patch on the Windows masses, literally hundreds of millions of computers throughout the world will compete for space to download it – and it is quite often very large: a 2005 estimate from Chris St. Amand of the Microsoft Operations team pegged Windows update traffic at 10-15 gigabits per second from the main download center during peak periods.

Comcast users represent a significant chunk of that number, being that the ISP is currently the #2 provider of Internet services in the United States. Where’s Comcast stepping in to manage its subscribers’ use of Windows Update?

The point I’m trying to make here is that BitTorrent isn’t the only bandwidth demon that haunts ISPs, and for Comcast and Congresswoman Bono Mack to accuse it of consuming “as much bandwidth as possible,” without mentioning many of the other applications that do as well, is obfuscating the real issue: that Internet consumption is expanding faster than ISPs would like it to.

Services like BitTorrent, and other P2P apps, have a legitimate place in today’s and tomorrow’s Internet. A number of progressive companies – Blizzard Entertainment, Joost, and others – use BitTorrent, or BitTorrent-like systems, to offload content away from the old guard of expensive, centralized client-server models and towards newer, P2P models. While this model does introduce an extra burden on the ISPs, I would think that ISPs should be responding to the needs of their customers, and not the other way around.

Food for thought: if Comcast invested all the money spent on excessive “network management” towards expanding the weak points of its infrastructure, how much further would the company be? How much further would the quality of our Internet service be? Instead of cracking down on protocols that Comcast deems unworthy, wouldn’t it be easier to promise – and enforce – a usable minimum of bandwidth to subscribers, and allow the rest of the pipe to be a free-for-all?

You can tell that Comcast is on the defensive. In the places where its response to the FCC isn’t littered with narcissistic praise, it constantly qualifies its “network management” practices with words like “little” or “minor” or “reasonable,” and there’s always an emphasis on the well-being of “all” users. Frequently these qualifiers appear italicized for emphasis. Don’t get me wrong, I’m all for reasonable network management – and indeed Comcast names many examples of reasonable network management in other companies – but none of those definitions include Comcast’s, who thinks that injecting forged disconnect packets into P2P sessions is “reasonable.”

Comcast contends that its practices are necessary so that it can stay viable in an intensely competitive – hah! – industry. I am officially calling shenanigans on this claim. I think Comcast is unwilling to further invest in its infrastructure’s weak points despite its considerable financial growth in 2007, and that it is unwilling to support an Internet that is growing faster than its master plan.

The Greek philosopher Heraclitus once wrote that “nothing endures but change.” Comcast, just like most any big company, is simply fighting change – and all the disruptions that come with it.

But we all know what happens when you fight change, right?

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

Sorry Tom, but..
By masher2 on 2/19/2008 12:09:33 PM , Rating: 3
> "There are many, many bandwidth hogs out there that are just as good, if not better, at “clogging the tubes” than BitTorrent and its ilk."

This just isn't true.According to some surveys, P2P data consumes up to 90% of all traffic on the Internet.

I don't support P2P throttling, but to claim its not the largest user of bandwidth on the Internet by far is flat out incorrect.

RE: Sorry Tom, but..
By KristopherKubicki on 2/19/2008 12:36:02 PM , Rating: 3
I've heard figures all over the board as far as P2P data. I've also heard that YouTube is 15% of the Internet traffic; that spam is 80% of all email; that MySpace serves more pageviews than Google and Yahoo combined.

(Tom works at an ISP BTW, at least for the next two days before he becomes at DT employee full time)

I've not seen concrete, accurate studies of traffic on the Internet. I don't suspect they exist. Even though we're a couple decades into it, the Internet is still in a state of flux anyway. What could be said to be true just three years ago is totally bogus today.

I think we need to ride it out a bit before proclaiming hard numbers in any direction.

RE: Sorry Tom, but..
By geddarkstorm on 2/19/2008 12:42:10 PM , Rating: 1
And don't forget on-line games! Those can be quite bandwidth intensive. What percentages do they accumulate to? Heck, what percentage of traffic is WoW alone?

RE: Sorry Tom, but..
By masher2 on 2/19/2008 12:49:50 PM , Rating: 2
Wow has only 8 million subscribers, the average who plays what, about 20 hours a week max? And the game itself consumes less than 128kb bandwidth.

Compared to the much larger P2P user base, consuming 100% of their 3mb, 6mb, or even 30mb pipe, often on a 24x7 basis, I'm sure you can see the two aren't comparable at all.

RE: Sorry Tom, but..
By geddarkstorm on 2/19/2008 1:00:12 PM , Rating: 2
I've used P2P for legitimate reasons (OC remix collections), and I've never achieved speeds, download or upload combined, that were even 2/3rds of 128kb/s on a 3mb/s connection. P2P seems extremely slow in my experience, although that could easily not be the norm, but I'd like to see actual figured on the matter before I'm swayed that P2P accounts for most traffic.

RE: Sorry Tom, but..
By masher2 on 2/19/2008 1:30:38 PM , Rating: 2
As my other post indicates, the estimate comes from a hardware vendor of traffic monitoring equipment; it's not a figure pulled out of a hat.

RE: Sorry Tom, but..
By Christopher1 on 2/19/2008 2:40:54 PM , Rating: 3
As my other post indicates, the estimate comes from a hardware vendor of traffic monitoring equipment; it's not a figure pulled out of a hat.

And you don't see the problem with that, in that you are taking the word of a company that has a FINANCIAL STAKE in selling equipment to 'monitor' excessive traffic as rote truth?
Stupid, to the extreme.

RE: Sorry Tom, but..
By masher2 on 2/19/2008 3:01:33 PM , Rating: 2
You make an excellent example of the human ability to rationalize away inconvenient data that doesn't fit your preconceptions.

IPogue has no "financial stake" in publicizing false statistics on total P2P traffic. In fact, just the opposite. Their livelihood depends on their ability to generate accurate figures on traffic breakdowns. False data would seriously jeopardize their product image.

They have absolutely nothing to gain from inflating one particular traffic source over another. Such a claim is-- as you yourself put it -- "stupid to the extreme".

RE: Sorry Tom, but..
By The0ne on 2/20/2008 1:07:36 PM , Rating: 2
He just doesn't want to believe anything that he didn't say for himself from ANYONE or ANY BODY. This should mean he shouldn't be voting, shouldn't be buying name brand items, eating at fine restuarants, etc. That's because we all know people do research and from that make their choices. But maybe that's too far out for him :)

RE: Sorry Tom, but..
By SandmanWN on 2/19/2008 3:10:36 PM , Rating: 2
Their company makes hardware that breaks down ALL traffic and creates an analysis of that information. They have no stake in what traffic gets revealed in the process. Their stake is only that the data is accurate.

Its the ISP that has a stake in the data itself, not the hardware monitor provider and their equipment.

RE: Sorry Tom, but..
By FITCamaro on 2/20/2008 1:45:29 PM , Rating: 2
MMOs are designed to use the least amount of bandwidth possible so that anyone can play. They come nowhere close to comparing to P2P or Bittorrent. I'd say you're probably using less than 100MB of bandwidth in an hour of play vs. about 1.8GB of bandwidth in an hour with a maxed out 4Mbps connection.

RE: Sorry Tom, but..
By masher2 on 2/19/2008 12:43:51 PM , Rating: 2
Peer-to-peer applications account for between 50 percent and 90 percent of overall Internet traffic, according to a survey this year by ipoque GmbH, a German vendor of traffic-management equipment.

I can't vouch for the accuracy of their figures, but given their hardware role, they'd certainly be in a position to know.

RE: Sorry Tom, but..
By IGoodwin on 2/19/2008 1:26:40 PM , Rating: 2
It is good that you posted the full quote about the statistic's range...

However, I hope this is not indicative of how you collect figures for your articles. A figure you have quoted with a 40% spread, which makes it obviously highly subjective, and you use the higher bound without qualification in an earlier post!

RE: Sorry Tom, but..
By masher2 on 2/19/2008 1:38:15 PM , Rating: 2
Without qualification? I specifically said "up to 90%" in my post, and further qualified it with "some surveys". I'd call that accurately qualified, especially given I was working from memory of a story read half a year earlier.

Point in fact, the range is primarily geographically based. The 50% low range comes from the Middle East (which accounts for a very low percentage of overall Internet traffic anyway). The high range comes from Europe. In Eastern Europe, it can go even higher still -- as high as 95% at times.

RE: Sorry Tom, but..
By IGoodwin on 2/19/2008 2:29:57 PM , Rating: 3
Lets just say we see your word usage highlighted the 90% figure without adequate qualification.

Anecdotal observation of peoples responses would indicate that the 'up to' part of your sentance was not sufficient to overshadow the 90% figure, leading to a skew in perception.

As I mentioned before, it is good that you do provide full quotes, as, in general, I like reading your material.

RE: Sorry Tom, but..
By geddarkstorm on 2/19/2008 12:38:57 PM , Rating: 2
That sounds totally incredulous. What evidence do they have to support that? I find it difficult to believe that most internet users would use P2P--or be using it all the time at absolute max settings with a client that's good enough to use all ones bandwidth--compared to how many users are just normally surfing the web. Not saying you're wrong at all, but I'd like to see credible studies on that, as 90% just seems like a ludicrous figure.

RE: Sorry Tom, but..
By masher2 on 2/19/2008 12:45:27 PM , Rating: 2
I doubt most uses do engage in P2P. But remember, a P2P app running 24x7 (as many do) can easily use 1,000X or more the bandwidth of a user who sporadically engages in email and web surfing.

RE: Sorry Tom, but..
By geddarkstorm on 2/19/2008 1:23:52 PM , Rating: 2
"Some have claimed the peer-to-peer file-sharing system BitTorrent accounts for one-third of Internet traffic. This estimate is far too high, but BitTorrent and its competitors are very substantial bandwidth consumers."

A very lengthy article, but full of information on the internet, its traffic, history and possible future. They seem to be legitimate and site a few peer review articles. It's the best real info I've found thus far.

RE: Sorry Tom, but..
By masher2 on 2/19/2008 2:17:49 PM , Rating: 2
Your link doesn't cite any reasons to discount those estimates. I'd put more faith in figures from actual hardware traffic analysis, than in an unattributed off-the-cost estimate from a academic talking head.

RE: Sorry Tom, but..
By Christopher1 on 2/19/2008 2:46:25 PM , Rating: 2
There is no evidence to support those numbers. Personally, even leaving a p2p applications like Shareaza or Azureus on for days straight, with it logging every single bit of data I transfer...... it usually comes to a grand total of 20GB's for 2 weeks straight.

That is less than I download while web surfing, according to the add-in that I found for IE7 that keeps track of every bit you download, which is anywhere from 30-40GB's, most of it demo games that I download.

The real problem today is that the IP companies have not kept up with the reality of distribution today, and have not realized that the people who are on the 'cutting edge' like myself (if I even am, considering that everyone who I ask admits to using Bittorrent/other p2p and streaming TV episodes from sites like are not going to be on the cutting edge for much longer.

RE: Sorry Tom, but..
By masher2 on 2/19/2008 3:05:00 PM , Rating: 2
On the contrary, there is very good evidence to support them:

See the pie chart for a complete breakdown by traffic type.

RE: Sorry Tom, but..
By geddarkstorm on 2/19/2008 3:27:44 PM , Rating: 2

Discounts them. "P2P applications now amount for 37 percent of the total traffic" according to Ellacoya Networks as of 2007. Youtube alone takes up 10%? That immediately disproves any figure of 90% for P2P, as there's more on the net than just Youtube.

Arbor networks also puts P2P at 37% .

The more I look, the more that 90% figure seems totally impossible. We now have three lines of evidence, two from industry and one from an "academic talking head" saying it's 37% or less.

RE: Sorry Tom, but..
By masher2 on 2/19/2008 3:42:19 PM , Rating: 2
> "We now have three lines of evidence, two from industry "

Err, Ellacoya Networks and Arbor Networks are the same company -- you just quoted the same survey twice.

So we have one survey which states P2P it is "50-90%" depending on area and another which states its 37%. However, they both agree that P2P is the largest consumer of bandwidth on the Internet, which was my original point.

In any case, I'd say the difference between the two surveys is most likely geographic distribution. Europe runs extremely high in P2P usage; North America is substantially lower.

RE: Sorry Tom, but..
By geddarkstorm on 2/19/2008 4:15:07 PM , Rating: 2
Ah, I did not know they were one in the same, so let me supplement with Cisco's 2007 study which says 50% in 2006 and losing ground to other activites to be 39% predicted in 2011,136069-page,1/ar...

Yes, it is obvious that P2P by itself has a sizable chunk that makes it the dominant single type of traffic. That is certainly not in dispute. But 50% seems to be the reasonable maximum that P2P takes up over the entire internet. 90% seems completely out of line, unless perhaps you only looked at one pocket, like a college campus that encouraged P2P, or something of that nature (as that seems the likeliest place for P2P to reach 90%, if it does anywhere on the net now a days). But this still calls into question the link you posted that makes the claim for P2P being 50 to 90% of "overall Internet traffic" (perhaps MSNBC took the study out of context). The revised estimate should be 37-50% then, working off only numbers we can be reasonably sure of. This certainly changes the gravity of the matter, though not the core issue of what single data stream is taking up the most resources.

In any case, the 90% of all internet traffic being P2P is definitely false (that has only been what I've been disputing), so Tom doesn't seem that far off with his statement, and the issue revolving around P2P isn't as grave as 90% would implicate.

RE: Sorry Tom, but..
By masher2 on 2/19/2008 4:35:48 PM , Rating: 2
Tom's statement was that there are many other applications "as good if not better" in consuming bandwidth than P2P. Even your own study claim otherwise. From your link:
Internet traffic is still dominated by peer-to-peer (P2P) applications and this trend is not slowing, the Cisco report finds . P2P traffic will more than quadruple from 649 petabytes per month in 2006 to 2,836 petabytes per month in 2011

> "In any case, the 90% of all internet traffic being P2P is definitely false "

As stated before, the figure was "up to" 90%. And in fact, apparently in some Eastern European countries, the figure can be even higher, rising up to 95% just from P2P.

RE: Sorry Tom, but..
By geddarkstorm on 2/19/2008 4:57:59 PM , Rating: 2
Again, I was never debating the point that P2P takes up the largest share for any one data steam type. Nor was I saying Tom was right. But, 90% for the over all internet is definitely incorrect, and by quite a wide margin. That's what the original "up to 90%" claim was referenced to in the actual link. And it is that alone that I was disputing.

It still seems ludicrous that even a single country would be as high as 90, let alone 95% in P2P (unless the Pirate Bay is now a country and I missed the news clip). That doesn't make it impossible, but I'd still like hard cold facts, as it has many implications in networking design and usage, and other interesting sociological connections.

RE: Sorry Tom, but..
By MatthiasF on 2/20/2008 10:21:03 AM , Rating: 2
I support P2P bandwidth throttling. The biggest reason, which seems to be left out in the debates, is the fact that BitTorrent and gnutella clients alike both try to make as many connections as allowed to download a file and use as much bandwidth as possible or allowed for each connection.

If you have a massive base that has no limits to the speed governing their constant traffic (assuming most clients are left open constantly to allow others to download) then they will always take up a significant chunk of the available bandwidth.

If today the average broadband line is 1 mbit upload and in a few years that doubles or tripples, so will the overall load being applied to the residential networks by these softwares.

Consider this, the average cost of a residential router for home is around $100. The average cost of a commercial router capible of DS3 speeds is over $10,000. DS3s can handle about 25 households at 2 mbits. At 60 million households, do the math. That's 2.4 Trillion US dollars so that you guys can share rediculious crap like music and videos.

Don't like the numbers? I was being conservative. Tweak them however you like, but you e-hippies are so far out of reality it's not funny. You'll never pay enough to support upgrades that extensive and yet you expect more and more.

The rest of us shouldn't need to pay for your bad habits. The net neutrality act was shot down and the freeloading mentality being spread (even by this blog post) is costing us more and more it seems.

RE: Sorry Tom, but..
By Methusela on 2/20/2008 3:25:44 PM , Rating: 2
Sorry, buddy, but you're wrong.

In general, ISPs oversell their bandwidth by a factor of 4-10. It used to be a factor of 12 or 15, back in the dialup days. This means that for every DS3 circuit, or 45mbit synchronous pipe, you most likely have between 90 and 225 people serviced.

Let's take the middle ground for our figures, or 160 people per 45mbit of bandwidth. At 60 million households, that's actually only $3,750,000,000 in capital expenditures for routing equipment. I think this estimate is still far too high, since this isn't a recurring expense.

In the end, that's just not a whole lot of money spread across so many different ISPs. Consider that it's only $62.50 per household, I'm just not convinced that you know what you're talking about, not to mention your spelling.

RE: Sorry Tom, but..
By MatthiasF on 2/20/2008 5:54:18 PM , Rating: 2
Oversold in the PAST, before the days of P2P. We're talking about what's necessary for the future.

If the mean (or worse the median) average for broadband usage increases substantially like has been hinted by not only all the statistics flying around but Comcast's actions of which you guys are complaining, they probably won't be overselling at all in the no-so-distance future.

Anyway, my calculation was way off. It should be $24 billion, not $2.4 trillion, for my scenario. Not sure where those two zeros came from in Excel (and of course I didn't save the file).

With that in mind, I'm changing my position. Comcast is a bunch of cheap bastards.

Verizon spends $15-18 billion a year upgrading their network, no reason why Comcast can't keep on par for their subscriber base (which is most likely smaller).

Oh, and on the spellcheck smack, I'm using IE6 today. I'm use to Firefox's form spellchecker. Hmmmm... can't wait for version 3.

RE: Sorry Tom, but..
By masher2 on 2/20/2008 8:41:43 PM , Rating: 1
Err, you're just counting the cost of the routers. That's peanuts. The real money is spent on the cables between those routers, as well as all the labor required to support and maintain the network. And it's not cheap.

For people who think it's is all lightness and air, fun and profits, remember that three of the top 10 largest corporate bankruptcies in history (including Worldcom, the very largest of them all) were all communications companies.

Maintaining a modern, national data network is *expensive*. If its not done just right, it costs much more than customers are willing to pay...and the company quickly loses billions, and folds up shop.

"My sex life is pretty good" -- Steve Jobs' random musings during the 2010 D8 conference

Copyright 2016 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki