Azureus Decentralizes Bittorrent

BobPaul writes "While the eXeem project to decentralize Bittorrent remains in open beta, the Azureus Java Bittorrent project has recently released a major update that, among other things offers 'a distributed, decentralised database that can be used to track decentralised torrents. This permits both "trackerless" torrents and the maintenance of swarms where the tracker has become unavailable or where the torrent was removed from the tracker.' It doesn't contain the search functionality of eXeem, but it's also not a beta product and is licensed under the GPL. Could this and compatible clients be the replacement to SuprNova and Lokitorrents, or does the lack of search negate its effectiveness?"

Comments

Torrent distribution

[Joakim A]

Torrents could be distributed in the swarms too. Possibly according to user preferences if the swarm has many torrents/many types of data. Could get really nice. We do need a python version though..

(Cant access the linked sites due to company policy (they allow /. :) so i don't know if this is supported.)

Re:Torrent distribution

[ratpack91]

you can run azureus headless and control it from the web interface- link

Re:Torrent distribution

[golgotha007]

ALthough Azureus is one of the better java written applications I've used, it still feels like java.

Everything updates just a little slow. You can be downloading via torrent something and have like 18%.
Then, go to another workspace and then later when I click back onto the Azureus workspace, it still shows 18% for about a second, then bam, all the values update. It's not my machine (p4,3.0ghz, gig ram).

Azureus, as an application, totally rules. It would be excellent in C or even python.

Re:Torrent distribution

[k98sven]

GNU Classpath and JamVM are smaller and faster than Suns JRE but they may not run Azureus yet.

No, JamVM (although extremly small ~200kB) isn't anywhere near as fast as the Sun JRE. JamVM is an interpreter, not a JIT VM. But it's fast as far as VMs go.

If you want a Classpath-based VM which is fast you should look at Jikes RVM or Kaffe, or perhaps consider compiling to native with GCJ.

Azureus uses native GUI widgets by way of the Eclipse SWT so if JamVM supports the required communication methods between VM and System alright, then it won't be too hard to run.

JamVM handles native calls without problems. I've run Eclipse and other SWT apps on it myself.

However, Azureus doesn't run on Classpath yet. It's very close to it.. But there's still one or two small issues with the Classpath libraries. If someone wants to help out with this, email the classpath list (classpath@gnu.org) or drop in on #classpath on FreeNode. You'll be 'liberating' Azureus and helping free java at the same time.

Kind of been done...

[Gilesx]

This is a little like Shareaza.

Shareaza has support for Gnutella, Gnutella 2, Edonkey and Bittorrent. As it provides a "bridge" between these networks, it means I am able to search for torrents from the two Gnutella networks, and edk. When I have this torrent, I can open it using the bittorrent part of Shareaza, and if that torrent is down, Shareaza will still hash the torrent and attempt to download the appropriate files from the Gnutella and eDonkey neworks. It's a nice idea, and really unites all the various p2p methods, using each method's strength to give an all round solid result.

I'm surprised that it's taken Azureus this long to catch up, and I'm sure we'll start to see a lot more bittorrent clients either offering their own solutions to this issue, or as in the case of Shareaza, using existing p2p networks to give backup to the Bittorrent protocol.

This is not anything new.

[solidox]

The bittorrent client BitComet has been doing this for a long time now.
Simply what it does is shares lists of peers between clients for matching infohashes...
It dosn't nessecerely decentralize it or remove the need for a tracker, as you need to get at least 1 ip from a member of the swarm (who has a compatible client)
It can help to get new peers if a tracker fails half way through, but you still need the initial peers ips from a tracker or similar.

Re:Tor

[moonbender]

Oh. The Tor guys won't like that. People figured out how to run BT over Tor a while ago (I tried it myself in January after seeing a presentation on Tor). Google for anonbt and you end up on a subsection of the Azureus homepage, saying:

Please *DO NOT* use Tor for routing peer-to-peer data traffic, it can not handle the bandwidth. They have indicated that they will make efforts to ban such usage if it continues, which will likely affect both legitimate and unwanted use!

Re:Using it now

[willdenniss]

It's UDP so there's no sockets anyway. Datagrams are connectionless and have far less overheads.

76,000 would still kill most apps though...

Will.

em..

[Stu Charlton]

Let's get some perspective. 266MHz isn't "a year or two" ago, it's been SEVEN YEARS since Intel released the P2 @ 266mhz. I have a 233MHz from that era, and you can barely even run Firefox on it (IE runs "OK"). Furthermore, 1997-1998 would be the era of Java 1.1 and 1.2, which were significantly slower.

These days, and since the year 2000 with the release of Java 1.3, Java UI's have been very usable. And Java is much faster than Python; it's comparing mixed mode dynamic compilation (Java) vs. interpreted (Python)! Pysco's JIT release in 2003 may have sped things up somewhat, but it's far from mainstream.

As for running on a 266Mhz machine, what's "plenty" of Python apps? Were they all graphical? I think you'd find graphical Python to be pretty pokey (pyGTK or what have you). Command-line Java is pretty fast.

Some statistics

[Saanvik]

Good point, you should have done it rather than posting about it.

Here are some statistics for your hungry little minds.

From the Illinois Council against Handgun Violence

In 2002, there were 30,242 gun deaths in the U.S.

Digging a little deeper, from the Department of Justice

The number of gunshot wounds from assaults treated in hospital emergency departments fell from 64,100 in 1993 to 39,400 in 1997, a 39% decline. Homicides committed with a firearm fell from 18,300 in 1993 to 13,300 in 1997, a 27% decline.

And from the Burlington Free Press

Vermont's loss of hunters is part of a national trend. The number of hunters declined from 14.06 million to 13.03 million, or 7.3 percent, from 1991 to 2001, according to the Census Bureau and the Fish and Wildlife Service. The drop was greater in the West -- 9.6 percent, from 2.46 million to 2.22 million.

Not a direct comparison, but it's hard to find numbers detailing the number of times a gun was discharged at a person versus discharged at an animal or target. Nevertheless, it's pretty apparent the original poster was incorrect. The vast majority of shooting in the US is not at people, but at animals and targets.

So, back on topic. The analogy was not a good one. A closer analogy could be made for handguns (handguns are not designed for hunting, but a lot of people do use them for target practice), but it still wouldn't be a good one.