Slashdot Mirror
BitTorrent Beats Kazaa In Traffic Numbers
prostoalex writes "CacheLogic attempted to measure the peer-to-peer network traffic by installing their network monitoring tools in data centers of large ISPs. The results are in, and Bram Cohen's BitTorrent overtook Kazaa's FastTrack network. BitTorrent traffic amounted to 53% of all peer-to-peer traffic, according to CacheLogic. It's worth noting, though, that Kazaa traffic is highly seasonal, as a lot of high-schoolers and college students are simply on vacation this time of year."
I remember on a NYT interview that Bram Cohen said that all BitTorrent packets were not encrypted nor decentralized. All machines must connect to the tracker in order to download, so there ARE ways to measure it.
They installed their monitoring system at the ISPs, so they can just analyze packets going into and out of the ISP's network. Kazaa packets and BitTorrent packets will look different and be destined for different ports, so it wouldn't be too difficult for the software to tell the difference.
"People that quote themselves in their signatures bother me" - athakur999
It's also worth noting that most people use BitTorrent for larger files like ISOs. So even though the traffic in bytes in higher, I'm willing to byte when it comes to "number of files transferred", Kazaa still has the lead.
"People that quote themselves in their signatures bother me" - athakur999
MPAA monitors bittorent traffic from sites such as suprnova.org. They constantly send out letters to ISPs that explains which movie was donwloaded, and how the ISP should proceed with the client. So, downloading several movies from suprnova.org is not a good idea, because MPAA sees what everyone downloads. BitTorrent is in no way an anonymous download.
Put up a freenet node and keep it running if you want to share things without having to worry about idiots on the network ruining it for you.
That said, I've never gotten freenet to work quite right and just recently removed it again in frustration.
Alito: A vote for Alito is a punch in the eye to put that bitch back in her place!
The quantity of data transferred is roughly the same, though. Perhaps as much as 1% higher due to protocol overhead.
Everything is explained in the documentation if you ever need to write your own client (or alternatively, you could look at the Python source code which is very interesting for all the lazy CS students on holiday like me ;)
"Shouldn't it work against both Kazaa and BT similarly? If kids are on vacation, they download less as a whole, so the decrease should be similar for both programs by my reckoning."
If we assume that kids have no interest in linux distros and other legitimate P2P uses and are dependent on P2P for copyrighted materials, then when kids are not online the majority of stuff that is still downloaded via P2P will be legitimate materials.
Since there are not many legitimate materials to download from kazaa, bittorent will come out ahead.
http://nyamenation.org/
Besides suprnova there is also: http://www.torrentreactor.net/torrents/ http://www.pleasure-torrent.com/ (porn only)
http://nyamenation.org/
Who needs P2P software when people leave movies unprotected on their websites all the time? Click on any website on this google search, see what movies they have, and leech em.
"'Yrch!' said Legolas, falling into his own tongue."
Well, I agree with your result - that the vast majority of bytes transferred using BT have the illegal bit set - but not with your means of getting there: looking at SuprNova is bound to result in a vast majority of the illegal stuff because, well, that's kind of what SuprNova is for. I mean, it's a (semi-) open tracker, sure, but I don't think a lot of people go to SuprNova for their legit torrents.
:)
Sites who do legit torrents usually have their own trackers, since setting up a tracker isn't a very large effort if you already have a site of your own and easily worth the control and overview it gets you. And on the other hand, individuals who do not have a site to spread torrents with rarely do legit torrents.
Of course this is all backed up by no evidence at all.
While I'm at it, there are several numbers that would be interesting to look at: The relative usage of the various P2P protocols - this is what TFA talks about. This is something you can probably determine fairly well by only looking at the port ranges involved. The percentage of legal traffic compared to the illegal traffic - ie what we've just been talking about. This is extremely difficult and most likely impossible to find out at the backbone level.
What I'm interested in is the percentage of the total bandwidth P2P makes up these days. Imagine if something like a third of the total ISP bandwidth is consumed by P2P file sharing - then consider that nearly all of it is illegal. At that point the currently practiced stance on copyright violation is just shown to be absurd: either DO something against it, enforce the laws you already have instead of inventing moronic new ones, or come to terms with a reality that probably won't go away any time soon.
Switch back to Slashdot's D1 system.
They do go after FTP servers. So does the MPAA. I know, I got a letter from them. They log in, grab a directory listing of things they believe are copyrighted and send the ISP a letter. Whole process probably takes 5 minutes. And for the record, their logging into sites listed on Oth.net. I'm sure their are others but I know Oth is being spyed on for sure.
My sig of choice is Marlboro
Now, this is just my view (which, surprisingly, might not have any basis on reality...), but i've observed that DC tends to be more "regional" and "communal" for a couple of reasons...
DC hubs don't scale well (at least the software that i know of), 1000-1500 users per hub seems to be the maximum, therefore hubs tend to be more private, usually, and anyone can start a hub. Plus most of the hubs i frequent have some restrictions on user population (for example, amount and types of files shared)
There's a couple of DC hubs in my local area that allow only local people to connect and people look for stuff there first, and then resort to other methods (bittorrent, ftp, kazaa etc.). The ISP knows of the existence of these hubs, but seems to look away, since it keeps the traffic to the outside world down. The pattern seems to pop up on many campus LANs as well. And yes, there's legal stuff in there as well (*gasp!*), a quick search shows the ISOs for all the major linux distros.
Maybe that's why... anyone can get to kazaa or use bittorrent, but dc hubs seem to be more restricted
because I'd bet over half the bandwidth on kazaa is people trying to re-download something. As opposed to bittorrent, where the quality of files is almost guaranteed.
All of that is great Dave, but Bearshare has spyware in it and it doesn't seem to stay up for me for more than a few days at a time. Why do you expend so much effort only to wrap it in such a crap program?
The tracker can be run on any port. Likewise, the clients can (and do) use any port. Many (and I mean a lot) of ISPs around the world rate limit BT traffic on the "well known ports". The only reliable way to get accurate numbers on bittorrent use is to decode the traffic -- peek at the actual content of each packet look for the BT protocol.
.torrent, it's a little difficult to identify what's being traded without piecing together all the data.)
Kazaa works much the same way.
(Without the actual
BitTorrent is rather unlike any of the other P2P systems out there.
From a user standpoint, it's "http, but vastly more scalable". It doesn't do file searching or anything like that. It just distributes load among all people downloading a file.
May we never see th
As if I'm going to install Bearshare and have my computer loaded down with spyware again. Fucking advertising drone.
-it has spyware
-its makers have tried their best to turn the gnutella network into a private playground
-apparently they are spammers as well
They don't have to inspect de packet content to determine what's the kind of traffic. They could use the stats that the routers gather.
l owscan/
I don't know how they did it, but here's how I would proceed:
You need is an access to the NetFlow data of a couple of Cisco routers on the ISP network. This data contains statistics about the connections between two IP adress like the port numbers, packets/bytes numbers and timestamps.
Then you setup a server to poll regularly the routers, aggregate the data and generate nice graphs using a software like FlowScan:
http://www.caida.org/tools/utilities/f
Nothing complicated, the difficult part is to explain to the ISP why they should enable NetFlow and give you access. (It adds a little load on the router).
I know that Autodesk recently converted there regression test farm to use bittorrent for transfers. It provides about a 10x speed increase. Autodesk is a BIG company, if the RIAA or MPAA tries to outlaw bittorrent, companies like this WILL fight back.
Not a sentence!
Not to mention they're pretty upfront about it. Guy's gotta make a few bucks somehow, and Bearshare has been rock solid (thanks Dave). Wouldn't use anything else for general P2P.
Linux, you magnificent bastard, I read the fucking manual!
For one, there's less load on the server(s). As long as the Tracker doesn't go under then mass popularity won't swamp the server, as quickly more and more of the downloaders can pick up the slack by offering up the chunks they already have.
If you can't get it down in one go, or something necessitates a reboot halfway through your Linux ISO set, it will always resume.
TiggsHeck, if the computer crashes then resuming will also check that what you have got isn't corrupt, and if it is then it'll automatically redownload the damaged parts.
Tiggs
"120 chars should be enough for everyone..."
Has he removed this obvious spyware in favor of a less intrusive money making model?
Agile Artisans