Felten on Comcast and BitTorrent

by on October 23, 2007 · 40 comments

Ed Felten isn’t impressed with Comcast’s traffic shaping techniques:

Comcast is using an unusual and nonstandard form of blocking. There are well-established mechanisms for dealing with traffic congestion on the Internet. Networks are supposed to respond to congestion by dropping packets; endpoint computers notice that their packets are being dropped and respond by slowing their transmissions, thus relieving the congestion. The idea sounds simple, but getting the details right, so that the endpoints slow down just enough but not too much, and the network responds quickly to changes in traffic level but doesn’t overreact, required some very clever, subtle engineering.

What Comcast is doing instead is to cut off connections by sending forged TCP Reset packets to the endpoints. Reset packets are supposed to be used by one endpoint to tell the other endpoint that an unexplained, unrecoverable error has occurred and therefore communication cannot continue. Comcast’s equipment (apparently made by a company called Sandvine) seems to send both endpoints a Reset packet, purporting to come from the other endpoint, which causes both endpoints to break the connection. Doing this is a violation of the TCP protocol, which has at least two ill effects: it bypasses TCP’s well-engineered mechanisms for handling congestion, and it erodes the usefulness of Reset packets as true indicators of error.

This brings to mind a question: as I understand it, TCP relies to some extent on clients being well-behaved and voluntarily backing off when faced with congestion problems. Is it possible that part of the reason that Comcast chose to target P2P applications specifically is that these aren’t “well-behaved” applications in this sense? Richard seems to be implying that this is the case. Is he right?

  • http://www.cato.org/people/harper.html Jim Harper

    Over at Harold Feld’s WetMachine, commenter Bill Clay writes:

    Perhaps as part of network neutrality, there should be an independent technical watchdog to regularly test and certify that ISPs conform to the relevant Internet technical standards (once upon a time this might have seemed appropriate role for the FCC). Even if non-conforming providers were not forced by any regulator to change their ways, a “Good Netkeeping” seal of approval might become a desirable ISP feature widely sought by informed netizens.

    This is a good idea, akin to Lauren Weinstein’s proposal to monitor for “neutrality” violations, which I applauded when introduced.

    (“Independent” to me does not imply government-run. Governments are very interested parties in the telecommunications space.

  • http://www.cato.org/people/harper.html Jim Harper

    Over at Harold Feld’s WetMachine, commenter Bill Clay writes:

    Perhaps as part of network neutrality, there should be an independent technical watchdog to regularly test and certify that ISPs conform to the relevant Internet technical standards (once upon a time this might have seemed appropriate role for the FCC). Even if non-conforming providers were not forced by any regulator to change their ways, a “Good Netkeeping” seal of approval might become a desirable ISP feature widely sought by informed netizens.

    This is a good idea, akin to Lauren Weinstein’s proposal to monitor for “neutrality” violations, which I applauded when introduced.

    (“Independent” to me does not imply government-run. Governments are very interested parties in the telecommunications space.

  • http://bennett.com/blog Richard Bennett

    OK, if we want to play watchdog, let’s do it right and include applications and users in the pool of abuse suspects.

    I guarantee you BitTorrent doesn’t conform to the TCP profiles in the RFCs, and nothing in the Internet’s traffic engineering lore deals effectively with bandwidth hogs.

    And how about monitoring Network Pundits for accuracy? Susan Crawford as much as accuses Comcast of identity theft, and that’s clearly incendiary and wrong. Take away her pundit’s license.

  • http://sethf.com/ Seth Finkelstein

    …target P2P applications specifically is that these aren’t “well-behaved” applications

    Whether they are well-behaved in terms of compliance with technical standards is rather besides the point. They are not well-behaved in terms of eating bandwidth, and that’s pretty simple.

    Regarding “watchdog”, been there, done that. See what happened with Global Warming report. Or TrustE. It’s just an invitation for propagandists to:

    1) Undertake a smear campaign against the accurate reports

    2) Create their own phony reports.

    The Libertarian bleat of “Government BAD, Business GOOD” is really tedious sometimes.

    Note this issue is essentially two business against each other – it’s not business vs. government

  • http://www.cato.org/people/harper.html Jim Harper

    Putting aside the unwelcome, flip portion of your comment, Seth . . .

    I don’t know what “Global Warming report” is, but one can cite Good Housekeeping and Underwriters Laboratories as counter-anecdotes. Obviously, some reputation programs work and some don’t. The idea is to have one that does, not to dismiss the baby with the bathwater.

  • http://bennett.com/blog Richard Bennett

    OK, if we want to play watchdog, let’s do it right and include applications and users in the pool of abuse suspects.

    I guarantee you BitTorrent doesn’t conform to the TCP profiles in the RFCs, and nothing in the Internet’s traffic engineering lore deals effectively with bandwidth hogs.

    And how about monitoring Network Pundits for accuracy? Susan Crawford as much as accuses Comcast of identity theft, and that’s clearly incendiary and wrong. Take away her pundit’s license.

  • http://sethf.com/ Seth Finkelstein

    …target P2P applications specifically is that these aren’t “well-behaved” applications

    Whether they are well-behaved in terms of compliance with technical standards is rather besides the point. They are not well-behaved in terms of eating bandwidth, and that’s pretty simple.

    Regarding “watchdog”, been there, done that. See what happened with Global Warming report. Or TrustE. It’s just an invitation for propagandists to:

    1) Undertake a smear campaign against the accurate reports

    2) Create their own phony reports.

    The Libertarian bleat of “Government BAD, Business GOOD” is really tedious sometimes.

    Note this issue is essentially two business against each other – it’s not business vs. government

  • http://www.cato.org/people/harper.html Jim Harper

    Putting aside the unwelcome, flip portion of your comment, Seth . . .

    I don’t know what “Global Warming report” is, but one can cite Good Housekeeping and Underwriters Laboratories as counter-anecdotes. Obviously, some reputation programs work and some don’t. The idea is to have one that does, not to dismiss the baby with the bathwater.

  • http://sethf.com/ Seth Finkelstein

    The “Intergovernmental Panel on Climate Change” – http://www.ipcc.ch/ (co-winner of the Nobel Peace Prize).

    Basically, big businesses which didn’t like IPCC’s research results simply generated a huge amount of noise to drown it out.

    Sort of like what’s happening here, actually.

    Underwriters Laboratories is IN EFFECT a public/private partnership between government and insurance companies, not exactly a Libertarian success story (almost the opposite, in terms of being a mixed-economy success story). Good Housekeeping is in effect an insurance company, it’s pretty interesting.

    Anyway, my point was that the idea doesn’t work because it has the same problem here in the first place – big businesses which don’t like the results simply trash any independent agency.

    Because what’s going with Comcast is not a tough call, at the core. Nothing to do with government – no business can afford to sell server-level bandwidth at home-use-level cost. No market competition will change that simple equation. It doesn’t take any Neutrality Commission to say that, yet making the point doesn’t do any good.

  • http://www.cato.org/people/harper.html Jim Harper

    Riiiiight. The IPCC – winner of the Nobel Prize – labors in corporate-manufactured obscurity. Exxon-Mobil planned the K-Fed/Spears child custody thing to keep the IPCC obscured . . . behind its Nobel Prize.

    But really. The IPCC is not a “seal” organization and thus can provide only the weakest, attenuated evidence against the viability of a good networking seal of approval. Yet you offer it up on the seal question along with sneers about libertarian ideas. If you feel misunderstood or ignored, try courtesy and tight, thoughtful argumentation for a change.

  • http://sethf.com/ Seth Finkelstein

    The “Intergovernmental Panel on Climate Change” – http://www.ipcc.ch/ (co-winner of the Nobel Peace Prize).

    Basically, big businesses which didn’t like IPCC’s research results simply generated a huge amount of noise to drown it out.

    Sort of like what’s happening here, actually.

    Underwriters Laboratories is IN EFFECT a public/private partnership between government and insurance companies, not exactly a Libertarian success story (almost the opposite, in terms of being a mixed-economy success story). Good Housekeeping is in effect an insurance company, it’s pretty interesting.

    Anyway, my point was that the idea doesn’t work because it has the same problem here in the first place – big businesses which don’t like the results simply trash any independent agency.

    Because what’s going with Comcast is not a tough call, at the core. Nothing to do with government – no business can afford to sell server-level bandwidth at home-use-level cost. No market competition will change that simple equation. It doesn’t take any Neutrality Commission to say that, yet making the point doesn’t do any good.

  • http://bennett.com/blog Richard Bennett

    Anyhow, I don’t think the average consumer gives a damn about anybody’s Geek Seal of Internet Purity; the issues are price, performance, and how soon you can hook it up.

  • http://sethf.com/ Seth Finkelstein

    There is no tight, thoughtful argumentation that will ever convince a Libertarian that his supposed free-market solution will be killed by malicious big business. It won’t happen. I do display a certain weariness on the topic.

    But, main point, again: No business can afford to sell server-level bandwidth at home-use-level cost. It doesn’t take any Neutrality Commission to say that, yet making the point doesn’t do any good.

  • http://www.cato.org/people/harper.html Jim Harper

    Riiiiight. The IPCC – winner of the Nobel Prize – labors in corporate-manufactured obscurity. Exxon-Mobil planned the K-Fed/Spears child custody thing to keep the IPCC obscured . . . behind its Nobel Prize.

    But really. The IPCC is not a “seal” organization and thus can provide only the weakest, attenuated evidence against the viability of a good networking seal of approval. Yet you offer it up on the seal question along with sneers about libertarian ideas. If you feel misunderstood or ignored, try courtesy and tight, thoughtful argumentation for a change.

  • Barsoap

    But, main point, again: No business can afford to sell server-level bandwidth at home-use-level cost. It doesn’t take any Neutrality Commission to say that, yet making the point doesn’t do any good.

    Ok, point taken. Now, let’s define ‘server-level bandwidth’, and while we’re at it – perhaps define ‘home-use-level’ as well.

    Is 64kb/sec upstream for an hour or two per day within ‘home-use-level’? Yes? No? Depends? Sometimes?

    It is easy to throw around opposing terms like ‘server-level’ and ‘home-use-level’ in order to make a convincing argument. But to define those terms, even generally or vaguely, is much less easy. If only Comcast would firm-up their definitions of terms such as those.

  • http://sethf.com/ Seth Finkelstein

    There are indeed hard cases. BitTorrent isn’t one of them. In fact, the exact opposite. BitTorrent is *designed* to use every bandwidth bit it can find, for file serving – that’s its goal in life.

    No definition which encompasses all possible situations need be developed here, because this is a very simple case indeed.

  • http://bennett.com/blog Richard Bennett

    Anyhow, I don’t think the average consumer gives a damn about anybody’s Geek Seal of Internet Purity; the issues are price, performance, and how soon you can hook it up.

  • http://sethf.com/ Seth Finkelstein

    There is no tight, thoughtful argumentation that will ever convince a Libertarian that his supposed free-market solution will be killed by malicious big business. It won’t happen. I do display a certain weariness on the topic.

    But, main point, again: No business can afford to sell server-level bandwidth at home-use-level cost. It doesn’t take any Neutrality Commission to say that, yet making the point doesn’t do any good.

  • Barsoap

    But, main point, again: No business can afford to sell server-level bandwidth at home-use-level cost. It doesn’t take any Neutrality Commission to say that, yet making the point doesn’t do any good.

    Ok, point taken. Now, let’s define ‘server-level bandwidth’, and while we’re at it – perhaps define ‘home-use-level’ as well.

    Is 64kb/sec upstream for an hour or two per day within ‘home-use-level’? Yes? No? Depends? Sometimes?

    It is easy to throw around opposing terms like ‘server-level’ and ‘home-use-level’ in order to make a convincing argument. But to define those terms, even generally or vaguely, is much less easy. If only Comcast would firm-up their definitions of terms such as those.

  • http://sethf.com/ Seth Finkelstein

    There are indeed hard cases. BitTorrent isn’t one of them. In fact, the exact opposite. BitTorrent is *designed* to use every bandwidth bit it can find, for file serving – that’s its goal in life.

    No definition which encompasses all possible situations need be developed here, because this is a very simple case indeed.

  • http://enigmafoundry.wordpress.com/ enigma_foundry

    First, Seth is right about the smear campaign against the IPCC report.

    To dig into the smear campaign a bit I would suggest two excellent sites:

    exxonsecrets.org & http://www.sourcewatch.org

    Second,

    Seth is also right about the informational exclusions that libertarians make–unable to ever recognize a market failure, or a successful government organization–they just can’t exist. Amartya Sen so cogently made this argument in Development as Freedom, I am tiring of repeating it here, as libertarians are without riposte to Sen’s arguments.

    A case in point is Jim’s recent argument made in a recent post, the response to which I shall post below:

    Once again, someone at TLF has defined a market success so narrowly, that by his very definition, it is impossible ever to discuss an example of a market failure. Here we have Jim Harper, discussing the recent supression by Comcast of Bit torrent traffic:

    “But I expect that we’ll soon learn more about the situation, and the conclusions to be drawn from it will be less obvious. There might be legitimate security reasons for what Comcast has done. We’ll see. We should expect full disclosure from Comcast.

    My take: If Comcast is “shaping” traffic inconsistent with their terms of service, for non-network-security reasons such as copyright protection or surreptitious usage control, they shouldn’t be doing that.

    More important is the meta-point: Independent testers found what they believe to be an impropriety in Comcast’s provision of broadband. They called it out, and interested parties among advocacy organizations and the media swarmed all over it. Comcast has to answer the charge, whether meritorious or not.

    These are market processes working their will, and the outcome will be reached in short order-”

    By this very low standard, it is impossible for there ever to be anything disclosed that is an example of a market failure and that would therefore require government intervention because if it is discovered and therefore discussed, Jim would just say something like:
    “My meta-point remains: Independent testing revealed alleged wrongful behavior by Comcast and an array of forces are requiring them to account for it. This is being done through operation of the market, without government intervention.”
    However, realize that this is just yet another example of a large corporation stifling public discussion to further its business plan. The internet is the new town square, and to permit toll booths and road blocks and secret protocols to intervene is unacceptable.

    Comcast has, despite the gnashing of teeth of the libertarians, convincingly made the argument for network neutrality legislation that no one else had as yet made so eloquently.

    The secret throttling of bandwidth is restraint of freedom of speech; many use Bit torrent to disseminate minority political speech that would otherwise be less accessible. Further, it cannot be ignored that the distribution of linux and FOSS to dismantle the centralized power structures of large corporations is itself an act with a political dimension. So, any attempt to dismantle or disrupt Bit torrent traffic is a de facto act of political repression.

  • http://enigmafoundry.wordpress.com eee_eff

    First, Seth is right about the smear campaign against the IPCC report.

    To dig into the smear campaign a bit I would suggest two excellent sites:

    exxonsecrets.org & http://www.sourcewatch.org

    Second,

    Seth is also right about the informational exclusions that libertarians make–unable to ever recognize a market failure, or a successful government organization–they just can’t exist. Amartya Sen so cogently made this argument in Development as Freedom, I am tiring of repeating it here, as libertarians are without riposte to Sen’s arguments.

    A case in point is Jim’s recent argument made in a recent post, the response to which I shall post below:

    Once again, someone at TLF has defined a market success so narrowly, that by his very definition, it is impossible ever to discuss an example of a market failure. Here we have Jim Harper, discussing the recent supression by Comcast of Bit torrent traffic:

    “But I expect that we’ll soon learn more about the situation, and the conclusions to be drawn from it will be less obvious. There might be legitimate security reasons for what Comcast has done. We’ll see. We should expect full disclosure from Comcast.

    My take: If Comcast is “shaping” traffic inconsistent with their terms of service, for non-network-security reasons such as copyright protection or surreptitious usage control, they shouldn’t be doing that.

    More important is the meta-point: Independent testers found what they believe to be an impropriety in Comcast’s provision of broadband. They called it out, and interested parties among advocacy organizations and the media swarmed all over it. Comcast has to answer the charge, whether meritorious or not.

    These are market processes working their will, and the outcome will be reached in short order-”

    By this very low standard, it is impossible for there ever to be anything disclosed that is an example of a market failure and that would therefore require government intervention because if it is discovered and therefore discussed, Jim would just say something like:
    “My meta-point remains: Independent testing revealed alleged wrongful behavior by Comcast and an array of forces are requiring them to account for it. This is being done through operation of the market, without government intervention.”
    However, realize that this is just yet another example of a large corporation stifling public discussion to further its business plan. The internet is the new town square, and to permit toll booths and road blocks and secret protocols to intervene is unacceptable.

    Comcast has, despite the gnashing of teeth of the libertarians, convincingly made the argument for network neutrality legislation that no one else had as yet made so eloquently.

    The secret throttling of bandwidth is restraint of freedom of speech; many use Bit torrent to disseminate minority political speech that would otherwise be less accessible. Further, it cannot be ignored that the distribution of linux and FOSS to dismantle the centralized power structures of large corporations is itself an act with a political dimension. So, any attempt to dismantle or disrupt Bit torrent traffic is a de facto act of political repression.

  • http://bennett.com/blog Richard Bennett

    enigma says: “The secret throttling of bandwidth is restraint of freedom of speech; many use Bit torrent to disseminate minority political speech that would otherwise be less accessible.”

    Sorry, but I have to call bullshit on you, dude. Who in the world depends on BitTorrent to communicate their political opinions? I can find everything from Nazis to Anarcho-Syndicalists on the Web, using plain old HTTP. Why I would need to use the proprietary BitTorrent protocol to get access to political speech is a total mystery.

    Or are you just trying to be hyper-dramatic?

    Seth makes the ultimate killer point here, by the way. Comcast will be glad to sell you a commercial account if you want to run a server, but the residential account is restricted from server operation. A BitTorrent seed is nothing more or less than a server, hence it’s a clear violation of TOS for residential accounts. If it’s all that important to you to serve up files with BitTorrent, buy a commercial account.

    This stuff really isn’t complicated, once you sweep the histrionics aside.

  • http://bennett.com/blog Richard Bennett

    enigma says: “The secret throttling of bandwidth is restraint of freedom of speech; many use Bit torrent to disseminate minority political speech that would otherwise be less accessible.”

    Sorry, but I have to call bullshit on you, dude. Who in the world depends on BitTorrent to communicate their political opinions? I can find everything from Nazis to Anarcho-Syndicalists on the Web, using plain old HTTP. Why I would need to use the proprietary BitTorrent protocol to get access to political speech is a total mystery.

    Or are you just trying to be hyper-dramatic?

    Seth makes the ultimate killer point here, by the way. Comcast will be glad to sell you a commercial account if you want to run a server, but the residential account is restricted from server operation. A BitTorrent seed is nothing more or less than a server, hence it’s a clear violation of TOS for residential accounts. If it’s all that important to you to serve up files with BitTorrent, buy a commercial account.

    This stuff really isn’t complicated, once you sweep the histrionics aside.

  • http://www.cato.org/people/harper.html Jim Harper

    Seth and enigma overplay their hands.

    Richard, you’ve brought a lot to this discussion. Thanks.

  • http://www.cato.org/people/harper.html Jim Harper

    Seth and enigma overplay their hands.

    Richard, you’ve brought a lot to this discussion. Thanks.

  • http://www.aleax.it Alex Martelli

    “TCP relies to some extent on clients being well-behaved and voluntarily backing off when faced with congestion problems. Is it possible that part of the reason that Comcast chose to target P2P applications specifically is that these aren’t “well-behaved” applications in this sense?”

    It’s technically possible for a good hacker to hack their Linux (or FreeBSD) computer’s TCP/IP stack so that it violates TCP standards and doesn’t back off properly on congestion. However, an application (p2p or otherwise) cannot possibly do that — it needs to be done by changing the operating system (if, like in all modern machines, the TCP/IP stack runs in kernel-land).

    If you successfully hacked your kernel’s stack, then all applications running on your machine would “benefit”… as long as no un-hacked network apparatus correctly implementing TCP was anywhere along your packets’ paths, of course (including the peer you’re communicating with, but also routers, firewalls, and the like, along the way). Targeting p2p specifically would be useless anyway (you’d just use other, non-p2p protocols and applications); using a watchdog to monitor leafnodes for such suspect behavior might work (but anybody who’s hacked their TCP/IP stack for such purposes surely knows enough to block RST packets too — so, Comcast’s behavior would NOT harm such uberhackers, only ordinary folks).

    So, summarizing: no, it is NOT possible that Comcast (who DOES have plenty of excellent network engineers, after all) is as confused as you are between the responsibilities of the various layers in a TCP/IP stack (and the extreme unlikelihood of such violations being widespread, in a world full of Windows machines, routers, etc, etc;-).

    Alex

  • jaminus

    Skype has 100 million users. Xbox Live has 6 million members as of March this year. iChat is growing rapidly. World of Warcraft uses Bittorrent to distribute game updates. These are very popular, residential-grade peer to peer uses. Don’t they constitute servers under your definition? They all accept unsolicited inbound connections using upstream bandwidth. Should Comcast hinder these applications as well? Surely you wouldn’t expect everyone who uses these services to sign up for business internet. Most people who use these servers don’t even rack up excessive bandwidth. These are commonplace residential internet uses and on an aggregate basis, peer to peer applications are used by a pretty sizable portion of broadband customers for legitimate reasons.

    Back when residential broadband was a newcomer, the client/server distinction made sense and residential users had no real need to use their connection as a server. As a remnant from the past, nowadays many providers still include a clause in their TOS very similar to Comcast’s ban on servers. But it’s widely understood that using residential internet as a server is acceptable as long as there are no sustained, commercial-grade high-bandwidth activities like web hosting. So even though most people are technically prohibited from using a server, many do without repercussion or so much as a peep from their provider.

    Comcast’s network can handle peer to peer traffic in moderation. Instead of simply banning these server-based protocols, Comcast would be wise to ensure residential users do not overutilize peer to peer applications. Simply educating people about the network congestion caused by peer to peer traffic and giving customers tips on “bandwidth conservation” would be steps in the right direction for Comcast.

    I am a Comcast customer because it is my only choice. Even though I chose my home largely because it was only 5000 feet from my CO and there are no technical limitations on my loop, Verizon refuses me DSL service because their CO has been maxed out for almost two months now. But my phone number qualifies on their system, so the only way I can check the status of my CO is to place a DSL order, pay $95, and wait two weeks to hear back from Verizon engineers as to the eligibility of my line.

    Comcast has given me zero incentive to shift my bandwidth-heavy tasks to off-peak hours. Nevertheless, I do my heavy downloading between midnight and 6AM, out of respect for their network. But most “hogs” are not so courteous. I download 720p movies from Xbox Live IPTV, and they are up to 10GB each. Just a dozen of these movies a month can generate some serious traffic. The only thing discouraging users from overutilizing their connection is the concern that Comcast will terminate their account for exceeding the secret cap. But it’s unlikely that Bittorrent uploading at 800 Kb/s will rack up the gigabytes fast enough to get an account blacklisted.

    Seth, your statement that “no business can afford to sell server-level bandwidth at home-use-level cost” doesn’t reflect reality. Given the current discussion is about Bittorrent, I assume you define server-level bandwidth as what’s used by peer to peer applications. But I challenge you to find a legitimate report of a FiOS or DSL customer being terminated or asked to upgrade service due to heavy P2P use. DSL and FiOS are business ventures which use aggregate usage to develop a price model that can generate profit without restricting bandwidth consumption or peer to peer traffic. DSL and FTTH do prohibit web hosting and block inbound servers on a handful of ports, but there’s no blanket server ban or blocking of peer to peer file sharing protocols. Obviously Verizon and AT&T use traffic shaping, but the impact of their throttling on peer to peer is far less noticeable than what Comcast uses.

    Comcast business-class broadband is horrible for residential use. Comcast workplace account representatives are not available except during weekday work hours. While residential service has 24/7/365 support for billing and technical issues, for whatever reason Comcast workplace cannot discuss account issues except Monday to Friday, 9 to 5. Plus the $250 installation fee is completely unjustifiable for residential users, because unlike businesses there is no need for Comcast technicians to wire a building for cable internet especially when all that needs to be done is replacing a residential grade modem with a business class one. And one-year mandatory contracts for workplace broadband are fine for businesses which have stability, but forcing customers to sign contracts is problematic because many people rent and switch homes regularly. Comcast’s triple play gives discounts to those who commit to a year of service but still lets customers elect against a contract, while business cable outright refuses any customer who declines a contract. Unlike cell-phone service, in many parts of the U.S,. Comcast isn’t the service provider, so if someone moves but they are in a contract they can’t just transfer service.

    If Comcast can’t make money because of peer to peer users, they would be smart to offer business-class service with residential power users in mind. Sure, it’ll cost $100 a month or more, but there is no doubt people are willing to pay, especially when Comcast is their only choice. No contracts or mandatory on-site installation should be necessary.

  • http://www.aleax.it Alex Martelli

    “TCP relies to some extent on clients being well-behaved and voluntarily backing off when faced with congestion problems. Is it possible that part of the reason that Comcast chose to target P2P applications specifically is that these aren’t “well-behaved” applications in this sense?”

    It’s technically possible for a good hacker to hack their Linux (or FreeBSD) computer’s TCP/IP stack so that it violates TCP standards and doesn’t back off properly on congestion. However, an application (p2p or otherwise) cannot possibly do that — it needs to be done by changing the operating system (if, like in all modern machines, the TCP/IP stack runs in kernel-land).

    If you successfully hacked your kernel’s stack, then all applications running on your machine would “benefit”… as long as no un-hacked network apparatus correctly implementing TCP was anywhere along your packets’ paths, of course (including the peer you’re communicating with, but also routers, firewalls, and the like, along the way). Targeting p2p specifically would be useless anyway (you’d just use other, non-p2p protocols and applications); using a watchdog to monitor leafnodes for such suspect behavior might work (but anybody who’s hacked their TCP/IP stack for such purposes surely knows enough to block RST packets too — so, Comcast’s behavior would NOT harm such uberhackers, only ordinary folks).

    So, summarizing: no, it is NOT possible that Comcast (who DOES have plenty of excellent network engineers, after all) is as confused as you are between the responsibilities of the various layers in a TCP/IP stack (and the extreme unlikelihood of such violations being widespread, in a world full of Windows machines, routers, etc, etc;-).

    Alex

  • jaminus

    Skype has 100 million users. Xbox Live has 6 million members as of March this year. iChat is growing rapidly. World of Warcraft uses Bittorrent to distribute game updates. These are very popular, residential-grade peer to peer uses. Don’t they constitute servers under your definition? They all accept unsolicited inbound connections using upstream bandwidth. Should Comcast hinder these applications as well? Surely you wouldn’t expect everyone who uses these services to sign up for business internet. Most people who use these servers don’t even rack up excessive bandwidth. These are commonplace residential internet uses and on an aggregate basis, peer to peer applications are used by a pretty sizable portion of broadband customers for legitimate reasons.

    Back when residential broadband was a newcomer, the client/server distinction made sense and residential users had no real need to use their connection as a server. As a remnant from the past, nowadays many providers still include a clause in their TOS very similar to Comcast’s ban on servers. But it’s widely understood that using residential internet as a server is acceptable as long as there are no sustained, commercial-grade high-bandwidth activities like web hosting. So even though most people are technically prohibited from using a server, many do without repercussion or so much as a peep from their provider.

    Comcast’s network can handle peer to peer traffic in moderation. Instead of simply banning these server-based protocols, Comcast would be wise to ensure residential users do not overutilize peer to peer applications. Simply educating people about the network congestion caused by peer to peer traffic and giving customers tips on “bandwidth conservation” would be steps in the right direction for Comcast.

    I am a Comcast customer because it is my only choice. Even though I chose my home largely because it was only 5000 feet from my CO and there are no technical limitations on my loop, Verizon refuses me DSL service because their CO has been maxed out for almost two months now. But my phone number qualifies on their system, so the only way I can check the status of my CO is to place a DSL order, pay $95, and wait two weeks to hear back from Verizon engineers as to the eligibility of my line.

    Comcast has given me zero incentive to shift my bandwidth-heavy tasks to off-peak hours. Nevertheless, I do my heavy downloading between midnight and 6AM, out of respect for their network. But most “hogs” are not so courteous. I download 720p movies from Xbox Live IPTV, and they are up to 10GB each. Just a dozen of these movies a month can generate some serious traffic. The only thing discouraging users from overutilizing their connection is the concern that Comcast will terminate their account for exceeding the secret cap. But it’s unlikely that Bittorrent uploading at 800 Kb/s will rack up the gigabytes fast enough to get an account blacklisted.

    Seth, your statement that “no business can afford to sell server-level bandwidth at home-use-level cost” doesn’t reflect reality. Given the current discussion is about Bittorrent, I assume you define server-level bandwidth as what’s used by peer to peer applications. But I challenge you to find a legitimate report of a FiOS or DSL customer being terminated or asked to upgrade service due to heavy P2P use. DSL and FiOS are business ventures which use aggregate usage to develop a price model that can generate profit without restricting bandwidth consumption or peer to peer traffic. DSL and FTTH do prohibit web hosting and block inbound servers on a handful of ports, but there’s no blanket server ban or blocking of peer to peer file sharing protocols. Obviously Verizon and AT&T; use traffic shaping, but the impact of their throttling on peer to peer is far less noticeable than what Comcast uses.

    Comcast business-class broadband is horrible for residential use. Comcast workplace account representatives are not available except during weekday work hours. While residential service has 24/7/365 support for billing and technical issues, for whatever reason Comcast workplace cannot discuss account issues except Monday to Friday, 9 to 5. Plus the $250 installation fee is completely unjustifiable for residential users, because unlike businesses there is no need for Comcast technicians to wire a building for cable internet especially when all that needs to be done is replacing a residential grade modem with a business class one. And one-year mandatory contracts for workplace broadband are fine for businesses which have stability, but forcing customers to sign contracts is problematic because many people rent and switch homes regularly. Comcast’s triple play gives discounts to those who commit to a year of service but still lets customers elect against a contract, while business cable outright refuses any customer who declines a contract. Unlike cell-phone service, in many parts of the U.S,. Comcast isn’t the service provider, so if someone moves but they are in a contract they can’t just transfer service.

    If Comcast can’t make money because of peer to peer users, they would be smart to offer business-class service with residential power users in mind. Sure, it’ll cost $100 a month or more, but there is no doubt people are willing to pay, especially when Comcast is their only choice. No contracts or mandatory on-site installation should be necessary.

  • Cheeseburger

    Would changing from a Comcast residential account to a business account somehow make upstream bandwidth any more available at peak times in residential neighborhoods? Are business-class accounts wholly exempt from the Sandvine filtering technology?

    If upstream bandwidth users are a money losing proposition, Comcast may just prefer that they go away. Much like all-you-can-eat buffets would prefer that gorgers not come around, or like how hospitals would prefer that those without insurance get sick somewhere else.

  • Cheeseburger

    Would changing from a Comcast residential account to a business account somehow make upstream bandwidth any more available at peak times in residential neighborhoods? Are business-class accounts wholly exempt from the Sandvine filtering technology?

    If upstream bandwidth users are a money losing proposition, Comcast may just prefer that they go away. Much like all-you-can-eat buffets would prefer that gorgers not come around, or like how hospitals would prefer that those without insurance get sick somewhere else.

  • http://bennett.com/blog Richard Bennett

    Folks, there’s a larger issue at work than you’re going to see with the “TCP relies to some extent on clients being well-behaved” mantra.

    The Internet as a whole, you see, depends on people using TCP in order to keep from going unstable. The entire traffic management capability of the Internet is wrapped-up inside TCP, and it’s quite easily defeated, without hacking any kernel code, simply by using capabilities already present on each and every system that connects to the Internet: it’s called UDP. UDP is a non-flow-controlled protocol intended for real-time applications where the TCP retry metrics aren’t appropriate, and BitTorrent uses UDP as well as TCP.

    Jaminus mentions some P2P applications, and lumps BT in with them, but that’s not the issue in this case. Comcast allows BitTorrent to run in P2P mode just fine (downloading while seeding.) They only get upset when your downloads are all done and you operate in seeding-only mode, which is, in effect, a server. Comcast isn’t down on P2P as far as I can tell, but it is very much down on Servers, and always has been.

  • http://bennett.com/blog Richard Bennett

    Folks, there’s a larger issue at work than you’re going to see with the “TCP relies to some extent on clients being well-behaved” mantra.

    The Internet as a whole, you see, depends on people using TCP in order to keep from going unstable. The entire traffic management capability of the Internet is wrapped-up inside TCP, and it’s quite easily defeated, without hacking any kernel code, simply by using capabilities already present on each and every system that connects to the Internet: it’s called UDP. UDP is a non-flow-controlled protocol intended for real-time applications where the TCP retry metrics aren’t appropriate, and BitTorrent uses UDP as well as TCP.

    Jaminus mentions some P2P applications, and lumps BT in with them, but that’s not the issue in this case. Comcast allows BitTorrent to run in P2P mode just fine (downloading while seeding.) They only get upset when your downloads are all done and you operate in seeding-only mode, which is, in effect, a server. Comcast isn’t down on P2P as far as I can tell, but it is very much down on Servers, and always has been.

  • http://enigmafoundry.wordpress.com e_f

    Richard Bennett, as usual, misquotes me and then takes apart his misquotation. First I said, in part:

    “The secret throttling of bandwidth is restraint of freedom of speech; many use Bit torrent to disseminate minority political speech that would otherwise be less accessible.”

    Then Richard Bennett says:

    Sorry, but I have to call bullshit on you, dude. Who in the world depends on BitTorrent to communicate their political opinions?

    I never said ‘depends’ but ‘use’, which is different.

    The books that I have in the past found using p2p include numerous books by Noam Chomsky. These are more available on p2p than in any other medium, although certainly if you live in NYC or Chicago you can find in a bookstore. But what if you live in Des Moines or Bismark, North Dakota?

    The second point I would make, that Richard Bennett ignores, and I believe is much more significant is this:

    Further, it cannot be ignored that the distribution of linux and FOSS to dismantle the centralized power structures of large corporations is itself an act with a political dimension. So, any attempt to dismantle or disrupt Bit torrent traffic is a de facto act of political repression.

    It is clear from Ed Felten’s work that this throttling was specific to Bit Torrent traffic, not just bandwidth. If an ISP wants to set a limit, so be it. That would be fine. I think most people want unmetered service, but if Comcast wants to offer a max throughput or max traffic per month that would also be fine, too.

    Just let us know what you are doing, so we can vote with our pocketbooks and go to another company, or enact regulation to stop it, if that is our democratic choice.

    Comcast did what they did in secret, and lied about it because they KNEW IT WAS WRONG.

  • http://enigmafoundry.wordpress.com eee_eff

    Richard Bennett, as usual, misquotes me and then takes apart his misquotation. First I said, in part:

    “The secret throttling of bandwidth is restraint of freedom of speech; many use Bit torrent to disseminate minority political speech that would otherwise be less accessible.”

    Then Richard Bennett says:

    Sorry, but I have to call bullshit on you, dude. Who in the world depends on BitTorrent to communicate their political opinions?

    I never said ‘depends’ but ‘use’, which is different.

    The books that I have in the past found using p2p include numerous books by Noam Chomsky. These are more available on p2p than in any other medium, although certainly if you live in NYC or Chicago you can find in a bookstore. But what if you live in Des Moines or Bismark, North Dakota?

    The second point I would make, that Richard Bennett ignores, and I believe is much more significant is this:

    Further, it cannot be ignored that the distribution of linux and FOSS to dismantle the centralized power structures of large corporations is itself an act with a political dimension. So, any attempt to dismantle or disrupt Bit torrent traffic is a de facto act of political repression.

    It is clear from Ed Felten’s work that this throttling was specific to Bit Torrent traffic, not just bandwidth. If an ISP wants to set a limit, so be it. That would be fine. I think most people want unmetered service, but if Comcast wants to offer a max throughput or max traffic per month that would also be fine, too.

    Just let us know what you are doing, so we can vote with our pocketbooks and go to another company, or enact regulation to stop it, if that is our democratic choice.

    Comcast did what they did in secret, and lied about it because they KNEW IT WAS WRONG.

  • http://bennett.com/blog Richard Bennett

    Last time I checked, E, Amazon delivers in Iowa.

  • http://bennett.com/blog Richard Bennett

    Last time I checked, E, Amazon delivers in Iowa.

  • jeena

    whether bittorrent provide both downloading and uploading?

  • jeena

    whether bittorrent provide both downloading and uploading?

Previous post:

Next post: