Net Neutrality and BitTorrent - No More Throttling? 243
Umaga's Purse writes "Will ISPs still be able to throttle BitTorrent traffic now that a significant proportion of it is legit? It's a tough question, especially for ISPs like AT&T (which agreed to run a neutral network in order to gain approval for its merger with BellSouth from the FCC). It's not just a problem for AT&T, though: 'ISPs that have made no such agreements may not need to worry about BitTorrent taking over their networks, but they do need to wrestle with the issue of how to handle it now that so many legal uses of the protocol are available. Do they want to irritate their BitTorrent-using contingent, or let BitTorrent flow unhindered at the risk degrading the experience of those who don't download torrents?'"
This may be a dumb question, but... (Score:4, Interesting)
So, if the tracker port number changes and the client port number changes, how is it being blocked?
Did I miss something? (Score:4, Interesting)
On what, exactly, are you basing this assumption that "a significant proportion" of BitTorrent traffic is legitimate?
Re:Which portion? (Score:3, Interesting)
But for the record, there were ALWAYS legit uses for BitTorrent. It's just that they're legitimate POPULAR uses now.
Trade off (Score:4, Interesting)
I guess the tricky part is at teh beginning when too big of a change may trigger a mass exodus. If they slowly start throttling it down and don't see much change in their business then they can keep that up until it becomes a problem.
Personally I think if/when ISPs do this they could avoid a lot of hassles by explaining it to people up front, in plain English, instead of burying their right to throttle your "unlimited" bandwidth in a cryptic and massive Acceptable Use Policy.
Re:This may be a dumb question, but... (Score:5, Interesting)
Classifying network traffic based only on the port went out the window well over 5 years ago when modern packet shapers came to the market which were able to analyze the very contents of packets and classify them based on the type of service they contained rather than the port they used.
Put in other words.. (Score:4, Interesting)
Remember this one? (Score:3, Interesting)
If Robert X. Cringely is right, then Google has indeed calculated well.
Re:Here's an idea (Score:5, Interesting)
How about before the ISPs even think of throttling down BitTorrent or any other type of traffic - they make even a casual effort to throttle back the 95% of email that is spam?
Why? Spam doesn't take up a significantly large portion of internet traffic and is a lot harder to separate out of the mix, than bittorrent. Even zombies performing DDoS attacks don't generally make up much of the overall internet traffic, although the spikes they create are problematic.
In reality, a number of large network operators don't want network neutrality. They want the opportunity to offer services and make sure competitors are unable to compete. They want to shake down companies individually by threatening to degrade their service and not their competitor's. They care about money; no hypocrisy there.
Re:This may be a dumb question, but... (Score:5, Interesting)
there are a number of ways, from deep packet inspection (studying packets and throttling those that appear BT-ish) to just cutting the uplink speed for a naughty subscriber. i think i my ISP may have done that to me already, judging by my ratios.
i do my own traffic shaping in my house with a linksys router running openwrt [openwrt.org] and x-wrt [x-wrt.org]. i do all my BT stuff from a vmware machine dedicated to all things BT (a win2k workstation running uTorrent [utorrent.com]) and i told the QOS config to file all traffic to and from his internal IP as bulk. i also use QOS to give priority to all traffic to and from my VOIP telephone adapter.
in case you are not a linksys firmware freak... putting openwrt on your router is like upgrading your PC to openBSD. loading x-wrt on your openwrt router is like installing KDE on your openBSD machine.
the result is BT can leech and seed 24x7x365, the humans in the house can surf and game unimpeeded and phone calls suffer no jitter from MMORPGS or BT.
i feel sort of like a hypocrite for being a net neutrality fanboy and using QOS inside my firewall... but at least i can trust myself to not degrade my access in favor of my own proprietary offerings.
some may say i am a little too trusting, but i have known me for a long time... i think we can trust eachother.
Re:This may be a dumb question, but... (Score:5, Interesting)
A sensible approach to make you happy (maybe) would be to limit the amount of bandwidth at each QoS level defined. If you want to burn your 500mb/month of highest QoS on bittorrent then so be it. Make the lowest tier of QoS truly unlimited. or some scheme like that.
Re:This may be a dumb question, but... (Score:3, Interesting)
My two cents says that it's none of my ISPs business what my packets contain. It may be their business how much bandwidth I use -- but it shouldn't matter if that bandwidth is VoIP, bittorrent, HTTP or a VPN. 100GB is 100GB regardless of what protocol generated the traffic.
Agreed, but net neutrality is about something more important than the type of traffic, it is the source of traffic. Large network operators have an interest in throttling traffic types, especially if they offer a VoIP service using one protocol and you're using another. They don't, however, need to know what is in your packets if they know the originating AS happens to be their competitor. They can just degrade all the packets to and from that AS that match the profile to insure their own offering is more reliable. They can threaten to slow traffic to any given web service and not their competitor unless that service provider pays up. In my opinion, stopping that is the important part of net neutrality, more so than packet contents, since we can and should all be moving to ubiquitous encryption anyway.
Re:This may be a dumb question, but... (Score:3, Interesting)
This may work in an ideal world, but the fact is that different applications do have different needs, and to make the Internet useful for more things it is necessary to have different levels of service
I do understand that point and I do use QoS on my own connection to prioritize SSH packets (need low latency) over HTTP/Bittorrent traffic. I guess my point though is that the ISPs should have enough capacity to provide low latancy (i.e: there shouldn't even be a queue) delivery to every packet. If they can't do that because 10% of the users are using 90% of the resources on bittorrent then they need to consider why they are selling unlimited service -- not what those 10% of the users are doing.
With the possible exception of a 911 call (or maybe Gov't/military operations) I'm not aware of a scheme in which users who pay more money get better access to the PSTN. If I obtain a timeslot/line/channel/what have you for my call then they don't get to tell me what I can do with it. If a timeslot isn't available then I have to try again later.
All that said we probably do need a balance somewhere between my ideas and those that would just throttle bittorrent until it becomes useless. Still, I hope I've made my point :)
Re:Value added (Score:3, Interesting)
My (possibly completely incorrect) impression of the problem ISPs have with BitTorrent is that it uses a lot of upload bandwidth at the last mile. Caching the data won't really help with that.
As I understand it, most ISPs have tons of bandwidth within their own network, but have much less bandwidth on the last mile. Essentially the last mile might be a 50Mbsp down/10Mbps up link shared among 20 customers. (Like 57% of all statistics, those numbers were made up.) So they might sell the connection as a 6Mbps/1Mbsp asynchronous connection to all those customers based on the typical web surfing usage pattern, where it's unlikely that any given customer will be using all of the bandwidth they're allocated.
If, instead, all of those 20 customers are participating in a BitTorrent swarm, they're completely saturating that last mile, and none of them can get the bandwidth they were sold. Worse still, if a mere 10 customers are able to flood the line, then the remaining 10 might actually get no access at all.
In this case, caching the data won't help - the ISP can't send and receive the data from their hub down to the customer line in the first place. Caching it might reduce the load on their backbone, but, as I understand it, that's not where BitTorrent overloads the network in the first place.
I know I have to keep my BitTorrent upload throttled to something like 50% of my max upload speed, or I can't do anything else, as BitTorrent overwhelms my available upload. Caching on the other end wouldn't help with that - I'd still be uploading enough to the local cache to overwhelm my own connection.
Re:Which portion? (Score:4, Interesting)
Blizzard undeniably uses bittorrent for the wow updater, yes, but me and all of my friends would argue the "quickly". It's dog slow and unreliable. No, its not a router issue or anything, we all torrent perfectly fine elsewhere (and if we were able to load the torrent in a good client like utorrent, maybe we wouldnt have a problem with this one). In the end a lot of people just close wow's uploader and wait until its up on fileplanet/filefront/etc.
I don't know whos fault it is, but I just had to throw that in there.
Re:This may be a dumb question, but... (Score:4, Interesting)
Using QoS on bittorrent is akin to my phone company telling me what I can discuss on the phone. In the end it should only matter how much bandwidth you use.
This isn't so, in general. QoS restricts traffic by type. So throttling bittorrent and prioritizing Web traffic is more like making sure regular voice on phones has priority over text messages, where that speed is less critical. The basic idea of QoS as it was initially conceived was to insure VoIP and video conferences did not lag, at the expense of a Web page loading a little more slowly or a bittorrent downloading a bit more slowly yet. This can be misused, say by degrading service on the ports used for one type of VoIP, and not on another, when your competitor offers their service on the one you're degrading. In general, however, encrypting packets makes this less important.
What is a real concern and needs to be addressed by net neutrality legislation is assigning quality of service that is different for the same traffic type, but for a different origin. Assume everyone moves to strongly encrypted packets and network operators have no idea what is in a given packet. That still doesn't stop them from assigning higher priority to packets that originate from their own VoIP servers and low priority to packets transiting their network from an origin that hosts their competitor's VoIP service. Worse yet, it does not stop some network operator who has no relationship with anyone but peering networks from going to Google and telling them all packets originating from Google's IPs are going to be set to a a lower priority than packets coming from MSN and Yahoo, unless Google is willing to pay an extra fee, and then going and doing the same thing to MSN, then Yahoo. Net neutrality with regard to who, rather than what, is a lot more important, in my opinion, than this focus on traffic types. I fear it is being overlooked in the discussion of this topic in the news and what that bodes for the resultant litigation.
Re:Here's an idea (Score:3, Interesting)
Regards QQ2
Re:Bandwidth Limiting (Score:4, Interesting)
I agree. Either give me exactly what I paid for (even if you have to adjust the price upwards), or advertise the REAL bandwidth (ie average connection speed), not some made up maximum theoretical speed if you're the only one on at 4:45 am. Overselling the service = selling something you don't have. That's tantamount to fraud if you advertise something you have no intention of providing.
Re:Which portion? (Score:3, Interesting)
Re:Which portion? (Score:3, Interesting)
Apply that throughout the whole torrent and the net effect is a slower rate of block distribution.
Blizzard have a constant HTTP seed providing blocks, so the torrent is never in an unseeded state, and at least one primary seed has a significant amount of bandwidth!
Incidentally, I think artificial throttling of protocols beyond the user's control to a given speed which is less than a reasonable "breathing room" overhead, or deprioritisation of a protocol to an extent that degrades performance, is an abuse of QoS techniques, and should probably be prohibited under net neutrality rules. QoS is meant to increase the overall performance of a user's connection, and of the network as a whole as a result.
Throttling should never be used as a method to increase contention or reduce costs - backend networks must continue to be expanded as traffic levels grow, or your whole branch of the internet will stagnate and be unable to keep up with growing demand.
The trouble with abusive QoS is that it invites the traffic classification rules to be evaded by protocol masquerading, scrambling and encryption techniques, so that the user can once again return to the reasonable level of performance they believe they are paying you for; and that means that other, beneficial QoS techniques can no longer be used on that protocol by you or anyone else.
In effect, abuse it and you lose it. Please don't abuse it.
QoS is a good technology when used appropriately to maximise the performance of a connection and network while minimising latency, but a bad technology when clamping everyone's torrent down to 5 bytes an hour because you "can't afford the bandwidth".
The Internet needs to continue to grow to maintain acceptable performance, and if you can't accommodate that at the rate you need to, your country will have a second-or-third-class backbone before you can even blink, and that's a competitive disadvantage for your whole country, and a national issue you need to deal with.
Re:Which portion? (Score:2, Interesting)
It doesn't appear limit the upstream like most bittorrent clients do, which means that your downstream gets throttled.
The best way to download the WoW patches is paste the
BitComet or uTorrent and let that handle the download instead. I find that the
download rate at least doubles that way.