Slashdot Mirror
AT&T Patents System To "Fast-Lane" File-Sharing Traffic
An anonymous reader writes Telecom giant AT&T has been awarded a patent for speeding up BitTorrent and other peer-to-peer traffic, and reducing the impact that these transactions have on the speed of its network. Unauthorized file-sharing generates thousands of petabytes of downloads every month, sparking considerable concern among the ISP community due to its detrimental effect on network speeds. AT&T and its Intellectual Property team has targeted the issue in a positive manner, and has appealed for the new patent to create a 'fast lane' for BitTorrent and other file-sharing traffic. As well as developing systems around the caching of local files, the ISP has proposed analyzing BitTorrent traffic to connect high-impact clients to peers who use fewer resources.
Seems like it would be easy for the marketing people to be singing their own praises while the core network people are quietly instructed to start using this software to catalog and ultimately curtail such practices.
I really would rather not have my ISP QoS anything that I do. I want them to be a common-carrier. I'll shape my own traffic, thanks.
Do not look into laser with remaining eye.
since there's nothing inside for them to inspect or cache.
(you are torrenting and NOT running a vpn? really? why?)
--
"It is now safe to switch off your computer."
Hard to swallow, but it violates net neutrality.
We supposedly dont want any preferential treatment of any traffic....
"His name was James Damore."
Anyone who uses this "fast lane" is going to be heavily targeted.
Lol, what ?
This system prefers a closer, better, faster HOST. Suppose your next door neighbor and a guy on the other side of the planet both offer a chunk of a torrent you want. It is better for it to be sent from your neighbor to you. That's faster for you and it's cheaper for the ISP than transporting traffic across the world or across the country. So that's what they patented - a system for encouraging your bittorrent client to download from your neighbor rather than from someone far away.
That's a preference for a particular host - the better one.
It wouldn't be a violation of net neutrality to completely squash all BitTorrent traffic.
When we talk about net neutrality, we're talking about treating the traffic the same regardless of source or destination. This is different from QoS where it's perfectly fine and useful to treat packets differently based on protocol. Yes, please slow down a web page load by a millisecond so a VoIP packet isn't dropped. One is noticeable, the other isn't.
And what's the source or destination of a request for (or seeding of) a BitTorrent file? The BitTorrent network. Doesn't matter which peer you're getting it from (on your end).
What this really is is a PR wedge in the door against net neutrality by making it seem like this is a net neutrality issue, when it isn't. "Just the tip...just for a minute..." "Oh, well I guess some fast lanes are okay..." And then they're giving it to you hard and deep.
We don't have a state-run media we have a media-run state.
Without reading the article (this is Slashdot, after all!), this sounds more like QoS management than creating a "fast lane". My cynical side tells me that they're calling it a "fast lane" just so that they can use it as an argument against the FCC in court.
"Gee, your honor. We were going to make our network faster and more efficient, but the mean old FCC said we couldn't put in any fast lanes! Government regulation is SO burdensome!"
Evil is as eval("does");
Go on, tell us what you REALLY think!
It wouldn't surprise me if when detecting bittorrent traffic AT&T disallowed connecting to any peers or seeds with an AT&T IP address. The downloader would still max out their up/downstream bandwidth, but it would be a single-edged sword as all their connected peers and seeds would be non-AT&T customers. AT&T would then have more available bandwidth (at the expense of the other ISPs) and could argue they were enhancing their customers' experience. A brilliant plan until other ISPs find out and do the same. Perhaps then AT&T could start their own VPN service marketed to their own customers under a different brand, touting unrestricted bittorrent connectivity as a selling point.
Buy your next Linux PC at eightvirtues.com
You're onto something, but coming at it from the wrong direction. CDNs do also (or at least can be used to) violate net neutrality. Security and convenience are always going to be fundamentally opposing ideals.
That will depend on implementation. Certainly most clients do prefer whoever's getting you the content the fastest. So if you've got a client set up to max out at 10 connections, it will either actively look for the 10 fastest or settle on the 10 fastest.
I didn't read the article - don't want Lumpy chiding me again ;) - but I'd almost think that AT&T would be planning on intentionally slowing down P2P traffic that goes outside of their own network (or whichever networks they pay higher fees on), making the clients automatically see the AT&T peers as being faster, and thus (mostly) selecting those anyway... even if they should have been slower.
I'm not even sure if that violates net neutrality unless each peer outside that network should be seen as competing with those inside that network.
From the comments, though, it seems more like they'd be trying to cache some bittorrent data and transparently serve that up to clients requesting it from what should have been a peer some hops down and/or redirecting requests from a particular peer (one that has high impact on the network) to another peer (one that has low impact on the network). The former should make a torrent download faster, the latter can go either way.
It's not super deep technical detail, but it's enough to be interesting. They're detecting BitTorrent traffic and pointing it to closer peers, so the traffic doesn't cross the network as many times, and doing file-sharing from some of their own servers. I couldn't tell from the article if the way they encouraged connections to closer peers was by adding delay to more distant peering connections, but that would actually speed up typical performance.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
Charge a hefty royalty for it.
Have gnu, will travel.
Get all the people who would be against it behind it by thinking they'll get faster downloads then once it's going either packet inspection and sell the infringers info to the xIAA, say 'psyche, net neutrality lol' or put an even faster fast lane for youtube/netflix etc traffic. For a premium of course. No way it will be as beneficial and sensible as the summary makes it seem.
Wanna buy a shirt?
https://www.redbubble.com/people/stealthfinger/shop?asc=u
(Disclaimer: I'm not speaking on behalf of any carrier, just speculating based on how Internet backbone and P2P technology work.)
That would fail badly. If Carrier A and Carrier B both did that, it would force most P2P peering connections to go through the network peering points between the carriers, which are just about the scarcest resource in carrier networks other than maybe cross-ocean or other international links. Each carrier would ideally want their own customers to do their P2P with each other, and do so at the nearest location (so users in the Northeast share with each other, users in the West share with each other, not too much crosses the long-haul network, YMMV about whether the US looks like 3 zones, 10, or 100.)
And if the carrier's doing their own P2P caching, they'd probably want to do some near the backbone peering points (e.g. San Jose, DC, maybe more) and encourage their customers to connect to those instead of running multiple streams across the peering point. Then you get to the engineering tradeoff question about whether you want a P2P cache in every cable head end, or just regionally.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
BitTorrent is something like 10%-20% of peak traffic, I'm not sure why they're so worried about it.
To get torrenters to oppose Net Neutrality and Title II legislation.
...ISPs do not care about file sharing much. People usually set a reasonable rate and download sporadically and through all times of the day. Most of these people don't want the data NOW RIGHT NOW and they aren't expecting live 4K video with zero stutter. Streaming content is a *BITCH* for ISPs because it's all lumped together at the same time of day - evening entertainment - and the customers want flawless video streaming at full rate in the highest resolutions possible. To multiple devices in the home, at the same time. And they do their best to convince the ISPs that they should support the home network and smart televisions, too (ISPs are having none of it, and for good reason).
They want it unlimited, as well. Free, if they could.
This really is a red herring. Streaming is what makes the Internet creak under the strain. It's essentially the worst possible, most expensive, most prone to issues way to distribute video to a wide market. Satellite broadcast is positively cheap, compared.
...Steve
A local host doesn't always mean "faster" for either bandwidth or latency. I've seen situations where others had lower pings to me with a different ISP than their neighbors on the same ISP, because of poor routing, and that's not including that they have highly asymmetrical down:up and I have symmetrical.
A simple change would be to make clients favor lower latency clients. Something like floor(log2(latency)), lower is better.
We all know AT&T is not actually trying to make P2P faster. They are really trying to:
1. Treat P2P traffic differently - probably slow it down or render it useless somehow.
2. Trick people into thinking this would be better than Net Neutrality.
3. Stick it to customers some other evil way.
Each chunk is downloaded from one peer. Also, normally from each peer you get one chunk at a time. So with two peers, the default behavior is to be downloading one chunk from the fast peer while downloading another from the slow peer.
That can often be less efficient than ignoring the slow peer(s) and just using the near/fast peer(s). If you and your neighbor are both AT&T customers with 30Mbps connections, the ideal is to transfer between you at the full 30Mbps. Using 5Mbps on the far peer with high latency and only 25Mbps on the local peer makes it slower for you and more expensive for the ISP.
"It's a trap!"
If it's about limiting consumer freedoms, I'm glad they're patenting it.
It exposes their ideas and it restricts others from doing it freely.
Because one of America's largest ISPs running DPI on BT traffic and building a cache of it to create a "fast lane" for traffic that they openly acknowledge is predominantly illegal transfers of copyrighted material would never be used for anything except making that predominantly-illegal traffic faster for users...