Slashdot Mirror
P2P Not Dead, Just Hiding
adavies42 writes "Contrary to media reports, P2P is not dying (PDF); it's just becoming harder to detect. In a paper for CAIDA, the Cooperative Association for Internet Data Analysis, researchers present evidence that the supposed decline in P2P traffic is actually due to a decline in easy-to-track protocols as those that change port numbers on a regular basis become more popular."
Tools such as Waste make this very easy to set up.
Not only is it not dying but it is being more and more adopted for non wzrez/moviez/pr0n related tasks. You can find pretty much any distro of unix/linux on p2p which takes a lot of strain off of the ftp sites. There is also a few schools i know of that use bittorrent to distribute movies that students create in class. So maybe the "P2P? Oh that is that software/movie piracy thing??" mindset is being corrected slowly but steadily?
TCP was designed with the assumption of a symmetric bandwidth path between the involved end points.
To try to put a figure on it, for around 80% to 90% of the Internet's history, the Internet has been run over symmetrical bandwidth links eg. 56Kbps full duplex point to point links, T1/E1s, T3/E3, Frame Relay, ATM, Token ring, the Ethernet variants etc. Asymmetric links such as DSL and cable are the exception.
TCP has performance issues when run over paths which involve asymmetric bandwidth links. They are described in RFC 3449 - TCP Performance Implications of Network Path Asymmetry.
The Internet's nature is peer to peer - 20050301_cs_profs.pdf
the likes of programs such as protowall and peerguardian, both of which have huge active communities constantly updating IP block lists, blocking all the p2p evils out there like bayTSP and other monitoring agencies.
A huge amount of p2p clients (most kazaa lite buids, azureus, one of the most popular bit torrent clients) have methods built in to support these block lists, and are turned on by default.
and are moving to decentralised networks like bittorrent
As much as bittorrent is the greatest thing since sliced bread, it is not decentralized. It is 1st gen P2P with a centralized tracker, despite actually being better than 2nd gen networks like KaZaA. And I certainly wouldn't want to compare it with 3rd gen networks like Freenet, MULE etc. which are at present even worse. So there's no shame in calling it 1st gen, far from it.
Of course, bittorrent more or less emulates a decentralized structure as each torrent operates independently of each other, but bittorrent itself is not. That does make it considerably harder to take down torrents than e.g. Napster, though.
Kjella
Live today, because you never know what tomorrow brings
From RFC 3449: "performance often degrades significantly because of imperfection and variability in the ACK feedback from the receiver to the sender."
prioritizing ACK packets with PF on OpenBSD
This was posted on Slashdot a while back. I personally use this for my home network with pretty much the same results posted there (as far as I can tell without doing formal testing). It's pretty much the only way I can keep my connection usable when my web server or torrents are having a good day.
PF is now available on OpenBSD, FreeBSD, NetBSD and DragonFlyBSD. I think some of them may still be lacking ALTQ support (needed to do the stuff in that link), but I'm not sure, I don't keep track of all of them.
I'm vaguely aware of NetFilter on Linux having similar capabilities, but AFAIK you need to tell it how to detect ACK and URG packets, it doesn't "just know" like PF does. I could be wrong about that, as it's been a long time since I looked into it. IPF and IPFW may or may not be able to do this, I have no knowledge of their capabilities in this area.
I rarely criticize things I don't care about.
With more and more people discovering the community enhanced joy that is bittorrent, I'm not surprised.
BitComet's default setting is to use a randomly generated port, and you can switch from port to port with the click of the "Random Port" button as often as you'd like.
Or you can choose to not listen on any ports, if you're like that, but you'll take a hit to the download speed.
MySpleen is one of the greatest torrent communities I've found, and if you're interested in MST3k, ATHF, Venture Bros, or the other Adult Swim 'toons, check us out!
Not a Twitter sockpuppet... but I wish I was.
No
The trouble is, you only have 2 real options, each with several implementations. You have Tor-like networks, which supplement their lack of content with outproxying to the internet. And then you have your freenet-like networks, completely internal (which I prefer ideologically).
The former seem to desire preserving the layer 3 protocols, meaning that they are (nearly) true networks that we are used to. However, even they have drawbacks... hidden services aren't currently able to have anything similar to domain names.
The latter, seem almost hellbent on being layer 7 protocols... and personally, I just can't see why people put up with this. It's not 1986 anymore, and even if it were... many BBS's had more functionality. Freenet in particular has only 2 functionalities that I'm aware of. Quasi-websites, and quasi-usenet (frost). How is this a solution for a project like bnetd? Even assuming that it's anonymity is strong, a major software project needs the infrastructure for people to coordinate. No email, no CVS... it just won't work.
And remember, even p2p is a dumb way to do what it was intended to do... napster wasn't invented because it's the ideal way to move mp3s... it was because even at the time, lawyers were spidering the web looking for someone's PWS website with MP3s. FTP's, when they could be found, had already been pushed underground by the anti-piracy efforts. P2P is the best effort solution against a nasty problem, not an ideal. Anyone with any sense, savviness or skill wants what p2p truly implies, that is, being a *peer* in the true networking sense of the word.
Allow me to describe another possibility. IPv4 itself allows for quite large private networks, 10net could concievably serve 16 million users. What we really need, is a way to string wires to each other, wires that are quite long, and not easily snooped on. We have that too. It goes by many names, but the category of software is called VPN. VPN software has many benefits, not the least of which it is deniable. If you're caught with freenet on your computer... you're using freenet. If you're caught with OpenVPN though, maybe you're trying to bypass how crappy WEP is, or connect to a work machine securely.
We need more than just a private network though, else the first narc who manages to get connected, can shut it down. We need a plausible way to be anonymous, and who can trust a mathematician? Your ass is *literally* on the line, if it doesn't work. The guys who do crypto are brilliant, but me, I want something I can understand without a PhD in number theory.
Well, first off, on this network, your internet IP needs to be exposed to as few people as possible. Freenet doesn't protect you from someone determining whether you participate. So, if we're passing packets around, we want that to be to as few hosts as possible. Second, we want to avoid creating any databases that correlate your 10.x.x.x address with an internet address or other identifying tidbit of info. That part is easy... routing IPv4 only requires knownledge of the geometry of the network, and which 10.x.x.x addresses are where within that geometry.
Take this for a small example of such a network.
A - B - C
A: 10.1.1.2
B: 10.1.1.3
C: 10.1.1.4
"A" can send packets to C, without knowing anything other than the 10.1.1.4 address. If the person in charge of A invites B, and tells B "invite someone else, and never tell me who it is", then A can't know who C is. And A only sends VPN tunnel packets to B, no one else.
There are ways to make this stronger though. Suppose all 3 hosts are in the same jurisdiction. When C innocently invites a narc (narcs are sneaky), the narc instantly knows C's identity. The narc is in the same jurisdiction and can easily get a subpoena or a search warrant on C's computer, on his ISP records, etc. B's identity is known quickly. B is also in the same jurisdiction... the same tactic will then reveal A's identity.
How do you protect against this? First, we insist
[Tips for running a successful Freenet node]
3 Most important ingredients:
Permanent connection
Bandwidth
Disk space
Without these you'll be complaining like the rest. Go ahead and set up a node, but optimum performance is a dream without all 3 above elements. Also, count on 2 days of letting it just run before you'll be able to get much done. After you're integrated things run much more smoothly!
If you're behind a firewall you'll need to know how to setup port forwarding. Windows install is the easiest, GNU systems should be trivial and there's a port for FreeBSD. I believe MacOSX can run it as well. If you can run a modern JavaVM, Freenet should be no trouble for you.
(About firewalls - if your $50 router/NAT/switch thingy cannot handle the hundreds of TCP connections Freenet can generate, you might want to either invest in a dedicated box (OpenBSD works well for me and allows me to prioritize traffic behind my interactive_ssh and vonage queues - Linux floppy distros should be fine too) or specify in freenet.conf to limit the number of open connections. Just be aware as connection tables can overload and distrupt the connection for all behind the NAT. Then again your $50 box may have no trouble at all. Port numbers are all random high port numbers making Freenet difficult to detect and firewall. Connections out will be made but the portforward is necessary for other nodes to connect to you. If nodes can't connect to you, performance will most likely be horrendous.)
If you just install Freenet and immediately try and download large files, you will be frusturated and give up. DON'T! Many freesites will not appear at all. NEVER FEAR! Let your node run in the background for a few days and get itself integrated into the mesh. Nodes that are more useful to the network (fast connection, large data store) will end up the most successful when downloading or uploading content. If you can't leave your machine running all the time or want to use freenet over dialup, fine, but your performance will not match those of others that can provide more to the network. Leeching is fine, it allows others to leech off of you - but leave your machine connected and Freenet's performance may end up suprising you.
Towards the beginning you may just want to start a number of downloads and count on many of them not completing - JUST WALK AWAY or do something else. Don't waste your time. By grabbing whatever bits you can, you'll increase the data in your own datastore and your connections within the network. If others find those bits from your node, your status will increase, more will connect to you and they will then be potential sources for more desired bits of your own. The better connectivity you've got, the more you will find. Leaving your node up at all times and keeping your datastore intact are the best ways to increase Freenet's performance (not just for you but for all).
THOSE PARANOID: I've been running my Freenet node wide open (no throttle) on my Earthlink cable connection in the heart of Raleigh, NC for some time. No threatening letters or trouble, my Vonage works fine (I do use pf's ALTQ) and those in my house have no trouble with connections, download or upload speeds)
For those that are already on Freenet and trying to download large files, one tool is critical. FUQUD (Freenet Utility for Queued Uploads and Downloads). Find it. Use it. Fred (the built in web interface) isn't going to cut it.
Regarding disk space. Unless you've got around say 2Gigs to dedicate to a node, your node may not perform as well as it could (200M is practical minimum). Consider the value you choose to be relatively permanant. You can't trade it with other uses - you build a datastore and that's the size, unusable for your MP3's or ogg's for example. They don't grow or shrink. You s
A good, and working, anonymous P2P alternative to Freent is I2P. The creator of I2P has been around for a while and cross talks with Freenet developers on occasion as both the Freenet and I2P community channels are on the anonymous irc network IIP, and irc.freenode.net.
A lot of I2P is put into the public domain, with parts of it being GPL. Try www.i2p.net for more information.
If the labels come after you for sharing such a file, tell them to sod off as the file in question is worthless without the other file which you did not share at the same time, did you?
The copyright merely has to "subsist in" the data, not be the recording itself. Even though worthless in isolation, I'm sure this would not hold water in court; it'd be an absurd interpretation of the purpose of the act if the only reason this file existed was to infringe copyright.
Incidentally, the careful choice of the words "subsist in" originally comes from, I do believe, copying large parts of others books into your own. It is ironic that it would apply on these points. Even if the words were less precise, the golden rule of statutory interpretation would protect the copyright owners: it'd be absurd to have for a user to choose to offer a file that contained data only used for copyright to be free of infringement by not having access to it on his own. The intention to infringe is there, as may be the intention to circumvent the law. At worst, it's contributing to infringement, probably exactly proportionate in liability to the amount that it is shared.
Now, if the user didn't intend to share the file, it's a whole different story. As well, there is a question as to whether the person sharing the a 'key' to the copyright infringed it, or contributed to its infringement.
Perhaps steganography would be more viable, being un-prosecutable by virtue of being undetectable.
They did actually, we got an MPAA letter about six months ago as the result of a download from suprnova.
I'm a signature virus. Please copy me to your signature so I can replicate.
Please, please, make sure the cryptography doesn't introduce too much load on the tracker. It's bad enough handling 2000 tcp connections/second with everything in plaintext...
Napster was the first real standalone p2p app, but the precursor to that was IRC channels dedicated to sharing. Much of the jargon that's used in modern p2p comes from that, for example, and the choice of encoding formats, e.g. mp3. Its almost certain that anyone who is at all serious about open source, hacking, gaming and any of a multitude of underground internet scenes has spent a good deal of time on an IRC channel at one time or another, and the scene is as much a social medium as it is a file trading medium.
Hell, if you want to go back even further, you could credit Doom and John Carmack with p2p. I haven't thought a whole lot about this, but it could be argued that the relatively open nature of Doom gave rise to networked group of file traders who would swap Doom mods and addons on BBS's, which were essentially prototypical networked IM and P2P applications (and occasional gaming platforms). The key to BBS's were that they were largely owner-operated -- you could chat with the owners, you knew them, they were part of the community. The notable thing about Doom filesharing was that creating Doom mods was a creative endeavor that benefitted from the free exchange of ideas. The runaway success of the scene spawned the idea that free and open trading of intellectural property was a moral good. Open source may have developed concurrently along the same lines, but I'm not sure there was very much cross-over. OS people were real coders and focused on that, and warez and mp3 people were more social and into gaming and things like that, almost like a collectors club. The intersection of these two groups in recent times greatly expanded on the ideas of the past, and with the addition of real coding ability came modern p2p, where coders no long limited themselves to the simple scripting environment of IRC clients, and went so far as to invent their own protocols and fully-fledged GUI environments designed to address the needs of specific internet sub-cultures. The user-friendly interfaces propelled them into the mainstream, e.g. Napster.
P2P enthusiasts are dead-serious about the importance of open intellectual property, and if its not seen as fundamental as Ghandi's struggle, it is seen as an important rights issue, which is what the parent was probably getting at more than suggesting that the P2P movement exactly follows the Ignore-Laugh-Fight-Accept model. Microsoft, RIAA, MPAA and the DMCA stifle human expression and creativity for profit and this causes immeasurable harm to society. The exact details of this harm are probably best left to another post, I may write a blog entry about it in the near future, but suffice it to say the lack of creativity and contribution in a person's life has a profoundly negative effect.
Ok, I should really go to bed...
"It's Dot Com!"
Actually, I got a letter for downloading but not finishing a movie a while ago. I started to torrent a movie (Black Hawk Down, for the record), and about 10 seconds and 20kb in went "Nah, don't really want to see it and its going too slow to even waste my time on" and canceled it. Month later, my ISP emails me saying they recieved a complaint about me sharing this movie. I was like "WTF!? I barely even downloaded or uploaded a sliver of it!".
:)
So no, they don't take the quantity of the infraction into consideration when mass mailing their letters. I suppose from their view, it makes no difference if you steal one car, fifty cars, or someones bumper: they still think you're guilty of GTA. Though your punishment if convicted would probably be different depending on the degree of your actions. Fortunatly for me, I live in Canada where the MPAA (or whatever the Canuck equivalent is) can bite my shiny metal ass.
Please explain how a stateful firewall could block BitTorrent-over-SSL while allowing HTTP-over-SSL to continue normally.
Mod down posts with a "Free Mac Mini/iPod" sig, they're spam!