BitTorrent Still Dominates Internet’s Upstream Traffic
torrentfreak.com
torrentfreak.com
Also, no mention of video chat. I believe google hangout, zoom.us, appear.in etc... work in peer to peer fashion when conferencing one on one, which is probably more difficult to measure, but I'm sure are significant contributor towards upstream usage.
This is really important because the upstream bandwidth offered by ISPs right now is ridiculous, and if the perception is that only "pirate" using BitTorrent use upstream, it's not going to get better any time soon.
Obviously this means "all HTTP(S) traffic we could not associate otherwise"
> Also, no mention of video chat. I believe google hangout, zoom.us, appear.in etc... work in peer to peer fashion when conferencing one on one, which is probably more difficult to measure, but I'm sure are significant contributor towards upstream usage.
Facetime is listed as 2.55%. I would guess all other video chat providers are way smaller.
> This is really important because the upstream bandwidth offered by ISPs right now is ridiculous, and if the perception is that only "pirate" using BitTorrent use upstream, it's not going to get better any time soon.
Well you could also say that over 70% of upstream bandwith is used for other protocols.
I see it as an indication that the current internet infrastructure is geared towards delivering content from a centralized instance rather than peer-to-peer services such as video chat, content sharing, streaming etc. Many consumer ISPs explicitly forbid serving content. It's a pity since some applications would benefit from symmetric lines. However, if everybody would be running their own home server we'd require a level of competence that's just not there in the general population - and probably never will since not everybody can be an IT expert. (just like I tend to rely on medical professionals if I need to).
In that respect, I find the notion of "X dominates the upstream" when X is one of the few applications where people currently use upstream on a constant basis pretty useless and unsurprising.
No, we'd just have to redesign the server software to stop assuming the operator has a high level of competence - just like Bittorrent servers already don't, in fact.
Oldschool P2p (ie Kazza and bearshare-style sharing schemes) are a different story, although even they are rarely attacked successfully.
Fyi, there are some tricks for overcoming bittorent encryption. I've used them to identify what was is shared across private networks (schools) but the techniques involve some trickery that is probably illegal on the public internet.
Assuming the traffic's being analyzed at a backbone level, it shouldn't matter that it's over VPNs, either -- it's still going to be BitTorrent traffic from some address, even if the "actual" source is unknown.
My usual client (transmission) actually picks a random port every time it starts. That does seem the norm for serious filesharers, which I also assume are responsible for a large percentage of torrent traffic.
Of course, whether or not something is copyrighted has very little to do with whether or not it is legal to distribute via BitTorrent.
I have been monitoring bittorrent traffic for 9 months, 24 hours a day, 7 days a week.
As of this second, there are 9,151,043 torrents in the system.
Out of the 9,151,043 torrents, 9,056,412 are pirated copies of either Software, Games, Music, Movies or TV. 98.9%
I'm maybe a little surprised that Linux ISOs account for so little. But then, I'm obviously out of touch.
Your answer was the sort of thing I was hoping for, although % of torrents may not be an accurate representation of share of traffic. Of course, it could very well be.
[1]: https://technifyme.wordpress.com/2013/02/26/netflix-uncovere...
You're sending a ton of real time data about stream quality and network congestion to help make sure you get the best possible experience. By adjusting your client to compensate for whatever local issues you might be having at that second and combining that with overall data, Netflix can do things like route around broken local routers and other sorts of things, as well as contact ISPs on your behalf if there are systemic issues.
That is awesome: So if I just continuously stream Netflix, I get free monitoring and incident response? Thank you! :)
More seriously, it's great that it works out this way. Most end users have no ability to identify the cause of problems much less do anything about them, and even technical users rarely have the time or resources (unless it's their job).
Since when? While most Bittorrent traffic uses encryption, but that encryption is not SSL, it does not use standard HTTP or HTTPS ports, in fact it often uses uTP which is not even TCP but UDP based. No bittorrent client I'm aware of makes any attempt to masquerade as HTTPS traffic. Some VPNs do, but that's about it. Bittorrent traffic using these protocols is trivially detectable and measurable, you just can't see inside of the connections (ie: know what content is being uploaded or downloaded) without doing an MITM, as they use a simple key exchange to prevent passive snooping, but you can see plenty of metadata, ports, protocol handshakes, etc to identify the protocol as a bittorrent protocol.
Just to be clear: there is an extension to make bittorrent use "encryption" (http://wiki.vuze.com/w/Message_Stream_Encryption), but while it does use crypto primitives its design make it closer to an "obfuscation" scheme than a real "encryption" scheme.
In the words of the author himself (https://torrentfreak.com/interview-with-bram-cohen-the-inven...):
> The so-called ‘encryption’ of BitTorrent traffic isn’t really encryption, it’s obfuscation. It provides no anonymity whatsoever, and only temporarily evades traffic shaping.