[00:00:06] *** Smirker has joined #bittorrent
[00:08:02] <TheSHAD0W> http://torrentfreak.com/researchers-improve-bittorrent-download-speeds-110706/
[00:08:30] <TheSHAD0W> Yeah, it's possible to come up with "superior sharing metrics" until you take gaming into account.
[00:18:31] *** K`Tetch has quit IRC
[00:25:58] <klapaucjusz> TheSHAD0W: PDF?
[00:26:14] * klapaucjusz is not going to read a paper in a Flash pager.
[00:26:24] <TheSHAD0W> It's not my paper...
[00:27:11] *** LordDoskias has joined #bittorrent
[00:27:26] <LordDoskias> hello - is there a decent bittorrent library in java?
[00:28:20] <klapaucjusz> TheSHAD0W: yeah, but you may have found the PDF.
[00:28:33] * TheSHAD0W sighs
[00:29:41] <klapaucjusz> Not on their web page.
[00:29:43] <klapaucjusz> Rand.
[00:29:45] <klapaucjusz> Rant.
[00:30:07] <klapaucjusz> Folks, how do you expect to be cited if you cannot be arsed to put your papers on your web page?
[00:30:23] <klapaucjusz> Folks, how do you expect to be cited if you cannot be arsed to put your final drafts on arXiv?
[00:30:51] <klapaucjusz> Folks, although I have full access to a good academic research library, I refuse to cite papers that are not available for free on the authors' web pages.
[00:30:57] <klapaucjusz> End of rant.
[00:32:54] *** Smirker has quit IRC
[00:33:28] *** Smirker has joined #bittorrent
[00:34:46] *** klapaucjusz has quit IRC
[00:58:33] *** freakazoid has left #bittorrent
[01:02:22] *** medecau has quit IRC
[01:02:52] *** _AdamK_ has joined #bittorrent
[01:36:50] *** ajaya has quit IRC
[01:37:56] *** K`Tetch has joined #bittorrent
[01:38:03] *** LordDoskias has left #bittorrent
[01:48:21] *** hydri_ has joined #bittorrent
[01:48:21] *** hydri has quit IRC
[01:48:22] *** hydri_ is now known as hydri
[01:50:28] *** multi-d has quit IRC
[01:57:36] *** swinokur has quit IRC
[01:59:54] *** Smirker has quit IRC
[02:00:23] *** Smirker has joined #bittorrent
[02:01:53] *** _AdamK_ has quit IRC
[02:15:43] *** swinokur has joined #bittorrent
[02:44:23] *** andar has quit IRC
[02:46:53] *** andar has joined #bittorrent
[03:08:44] *** multi-d has joined #bittorrent
[03:22:37] *** The_8472 has quit IRC
[03:32:55] *** Gottaname|Mobili has joined #bittorrent
[04:23:41] *** Gottaname|Mobili has quit IRC
[04:24:33] *** Gottaname|Mobili has joined #bittorrent
[04:42:50] *** btcod has joined #bittorrent
[04:43:22] <btcod> dear all, does any standard bittorrent client implement local neighbor selection (ie., favors neighbors in same AS)?
[04:44:03] <btcod> i wonder the pros and cons of such a strategy
[04:48:04] <alus> btcod: it was tried. in some popular situations (Comcast) it is a terrible idea
[04:48:52] <DeHackEd> I tried it in a tracker implementation...
[04:49:02] <alus> btcod: so it was determined that this decision should be up to the ISP, and several mechanisms were constructed to let ISPs do that, but none of them did
[04:49:31] <btcod> the tracker implementation involves favoring local peers when distributing neighbors?
[04:49:57] <DeHackEd> yeah. but my implementation was intentionally stingy about how much I give..
[04:50:18] <DeHackEd> 20% local, 20% proximity (if I can find them), and the rest random.
[04:50:27] <btcod> for Comcast it is a terrible idea?  but Comcast as any other isp would prefer to reduce its inter-ISP traffic, right?
[04:50:49] <DeHackEd> I mean, users will only upload to a handful of users, but open dozens of connections. I need to keep the diversity up
[04:50:53] <btcod> DeHackEd: how did you define local and proximity?
[04:51:03] <DeHackEd> routeview database
[04:51:06] <alus> btcod: inter-ISP traffic is not as much of a problem for them as local congestion. favoring local peers increases local traffic, and thus local congestion
[04:51:16] <btcod> DeHackEd: and do you know why it is not that popular?
[04:51:24] <DeHackEd> huh?
[04:51:28] <alus> btcod: also imagine a cell network. local traffic is a terrible idea.
[04:51:51] <btcod> alus: very interesting, local congestion is worst then inter-AS traffic? how so?
[04:52:27] <btcod> alus: do you think that nowadays BT is being used in cell networks?
[04:52:39] <DeHackEd> consider this: the upstream to the internet will be gigabits or tens of gigabits each. a CMTS will usually have a capacity of much less.
[04:52:57] <btcod> DeHackEd: what is CMTS?
[04:53:02] <DeHackEd> cable modem termination system
[04:53:04] <alus> btcod: local congestions affects the ability for other people to web-browse normally. local capacity is not infinite, and is not something which is easy to pay more to get a greater amount of. inter-AS traffic is
[04:53:05] <DeHackEd> the other end of the cable modem
[04:53:20] <btcod> CMTS are used to connected ISPs?
[04:53:27] <DeHackEd> no. it services cable modems
[04:53:27] <alus> btcod: I don't know how much BT occurs over cell, but it's an example of edge-congestion being the real issue
[04:53:54] <DeHackEd> a CMTS has two channels: upstream and downstream (always described as relative to the cable modem itself)
[04:53:58] <btcod> i'm wondering how prevalent is the edge-congestion problem as opposed to internal congestion
[04:54:28] <btcod> and how could I possibly quantify that
[04:54:29] <alus> btcod: quite prevalant. the internal capacity of the internet is effectively infinite compared to the edge
[04:55:06] <DeHackEd> upstream gets congested worst because one channel services several modems which need to do the whole CDMA thing
[04:55:12] <btcod> alus: so, do you think that some day people might prose that the bias in the neighbor selection is towards non-local nodes? :)
[04:55:24] <DeHackEd> upgrading frequencies isn't quite as easy as upgrading your contract with Level3
[04:55:49] <DeHackEd> also upstream channels are slower than downstream channels
[04:55:52] <alus> btcod: well, just non-local isn't the right answer either. I don't want a Comcast peer, I want a FiOS peer ;)
[04:56:12] <btcod> DeHackEd: when you say "upstream gets congested worst", you mean that the edge is the bottleneck, not the local network?
[04:56:34] <btcod> what is FiOS? :D
[04:56:47] <alus> btcod: and so biasing towards the fastest ASNs was considered, but then you need reliable centralized knowledge of that. randomly picking IPs and paying attention to which is fastest seems to be a pretty reasonable strategy
[04:56:56] <DeHackEd> between the cable modem and CMTS, there is congestion
[04:57:00] <alus> btcod: fiber to the home
[04:57:20] <DeHackEd> Verizon right?
[04:57:22] <alus> yes
[04:57:47] <btcod> well, picking fastest is the best from the perspective of the peers...
[04:57:54] <btcod> i mean, to increase their download rates
[04:58:04] <btcod> i'm wondering if this could also be best for ISPs
[04:58:05] <DeHackEd> pfft
[04:58:15] <DeHackEd> if 100 users all offered to let you download from them, you'd take it. no exceptions
[04:58:49] <btcod> given the above argument, that it is best to pay to use an edge-link rather than to have congestion in your internal network
[04:58:54] <alus> btcod: as far as I can tell, "best" for ISPs varies by ISP. more concerningly, they won't admit what their pain points are or how to fix them, because they think that is valuable competitive information
[04:59:11] <DeHackEd> ADSL would have it better.
[04:59:28] * DeHackEd works for an ISP who offers both DSL and cable modem services. So he knows a bit of the backend functionality
[04:59:39] <btcod> i understand that this is valuable competitive information
[05:00:19] <btcod> so, the idea of biasing traffic to avoid inter-AS connections is broken?
[05:00:31] <DeHackEd> not broken. but it's not the end-all solution
[05:00:44] <btcod> i see
[05:01:55] <btcod> DeHackEd: what do u mean about knowing the backend functionality?
[05:02:40] <DeHackEd> I mean I've made cable modem config files. I've seen the technical config of a CMTS. Most users don't know what's on the other end of a cable modem
[05:03:07] <btcod> i see
[05:03:33] <btcod> so, if you were running an ISP and you could set up your own tracker, you would not favor always the internal connections?
[05:03:44] <alus> DeHackEd: TheGoogle is on the other end. everyone knows that
[05:03:58] <DeHackEd> alus: correction: TheGoogle knows that
[05:04:02] <alus> indeed
[05:04:10] <DeHackEd> http://en.wikipedia.org/wiki/Cable_modem_termination_system
[05:04:57] <btcod> thanks
[05:05:10] <DeHackEd> btcod: I know enough about bittorrent that I would use the implementation I described in my little test tracker.
[05:05:32] <DeHackEd> a bias towards local, but not so much of a huge one that your connection diversity is compromised
[05:05:59] <btcod> what about the potential problem of overloading the internal network?
[05:06:18] <btcod> you would not worry about that?
[05:07:03] <DeHackEd> those same local connections will be carrying a reasonably substantial amount of my outer-internet traffic around my network. Shifting that traffic inwards won't seriously affect my internal wiring
[05:07:05] <btcod> i still don't get what is easier/cheaper.... to upgrade the outbound connection or to upgrade the internal network
[05:08:15] <btcod> i understand that if the internal network is a cellular network, it might be cheaper to upgrade the outbound connection, right?
[05:08:22] <btcod> but is this an exception?
[05:08:22] <DeHackEd> it depends. upgrading the outbound connetion depends on what my hardware is and what my contract is. For example, if I'm a small ISP and only have a 50 megabit connection to the internet (let's say I do dialup) it's probably implemented as plain Fast Ethernet and rate-limited. I could upgrade to 100 by simply removing the rate-limit
[05:08:52] <DeHackEd> okay... I'm going to define 3 segments:  out to the internet, my internal routing, and me->customer connections.
[05:09:10] <DeHackEd> in the case of a cell company, out to the internet connects to big companies like Level3, cogent, etc
[05:09:20] <DeHackEd> my internal routing is fiber in the ground I laid down or lease.
[05:09:34] <DeHackEd> and me->customer is from the cell towers to the cellphones themselves (wireless)
[05:09:42] <btcod> ok
[05:10:03] *** goussx has quit IRC
[05:10:27] <DeHackEd> upgrading my level3 connection could be as easy as a contractual upgrade. then level3 unlocks some speed on my rate-limited 1-Gigabit connection to them
[05:10:58] <DeHackEd> to exceed 1 gigabit, I need to either upgrade the wiring to 10-gigabit, or maybe run another wire so I could now reach 2 gigabit... annoying but probably not too difficult.
[05:11:03] <DeHackEd> these are just numbers
[05:11:14] <btcod> i got it
[05:11:32] <btcod> on the other hand, upgrading me->customers would be the most difficult change in this case, right?
[05:11:33] <DeHackEd> my internal wiring is much the same. if I have my own fiber, I can replace a 1-gigabit receiver with a 10-gigabit on each side and get 10 times the capacity...
[05:12:01] <DeHackEd> right. in the case of cellular you have set up new towers, or upgrade towers to support more users. I'm not a cellular expert but I expect you'd need more radios with more frequency coverage.
[05:12:18] <btcod> interesting
[05:12:21] <DeHackEd> and frequencies are limited by bodies like the FCC. and cell phones radios will have limits
[05:13:17] <DeHackEd> cable modems do operate in a similar manner. downstream and upstream are distinct channels. downstream is faster, and has the advantage that only the CMTS needs to transmit on it, so there isn't the risk of transmission collisions
[05:13:37] <DeHackEd> upstream is slower, and there's issues if two modems transmit at the same time.
[05:13:57] <btcod> correct me if i'm wrong, but all this means that if bt users are in a cell network, it might pretty much be that it is best to prioritize outbound connections rather than internal ones...
[05:14:32] <DeHackEd> I think the issue is that the wireless bandwidth will fall long before the outbound connection falls. Therefore it's not the outbound connection you should be worried about.
[05:15:12] <btcod> i see
[05:15:29] <DeHackEd> by fall, I mean fall over and die
[05:15:43] <btcod> the wireless bandwidth will die before the outbound
[05:16:23] <DeHackEd> this is certainly true in the shared-medium connections. cable modems and all wireless technologies qualify
[05:17:20] <DeHackEd> and to a certain degree, internal ISP connections (type "2") may fall victim to the same thing if you do something strange, like get all users in city A to bittorrent to users in city B, even though they're all from the same ISP
[05:17:23] <btcod> so, it is always best to prioritize local connections ... if local connections die after this prioritization, it means that the system can't support the demand anyways, and there is no way to fix that by moving traffic that is internal to the outbound link
[05:18:39] <DeHackEd> bittorrent by nature searches for peers to maximize transfer rates. assuming you have a good peer selection (both local and remote) and otherwise normal behaviour you'll hunt around for better options. if you find them externally, you'll just saturate the wireless network even harder.
[05:18:53] <btcod> if the internal traffic of bittorrent users in city A is depleting city A connection, shifting the traffic to city B will not address the problem
[05:19:07] <DeHackEd> uhh... I think I lost you
[05:20:16] <btcod> i think your point is that in any case if the internal connections are saturated, there is no way to address the problem by using external links, right?
[05:20:34] <DeHackEd> well, the typical user only has one internet uplink, right?
[05:20:43] <btcod> yes
[05:20:45] *** goussx has joined #bittorrent
[05:20:49] *** goussx has joined #bittorrent
[05:20:50] <DeHackEd> one way or another, that cable modem is the only way out
[05:20:58] <btcod> right
[05:23:41] <DeHackEd> well, if there's nothing else I'm going to bed.
[05:23:53] <btcod> thanks!
[05:24:20] *** Kitsoran has quit IRC
[05:24:43] *** niekie has quit IRC
[05:27:12] *** niekie has joined #bittorrent
[05:40:44] *** goussx has quit IRC
[06:08:49] *** goussx has joined #bittorrent
[06:10:12] *** goussx_ has joined #bittorrent
[06:10:12] *** goussx has quit IRC
[06:10:12] *** goussx_ is now known as goussx
[06:55:53] *** MassaRoddel has quit IRC
[07:12:38] *** MassaRoddel has joined #bittorrent
[08:02:04] *** btcod has quit IRC
[08:29:32] *** Smirker_ has joined #bittorrent
[08:32:10] *** Smirker has quit IRC
[08:59:33] *** Astro has quit IRC
[09:03:20] *** Astro has joined #bittorrent
[10:02:36] *** gde33 has quit IRC
[10:27:26] *** whiteybulger has quit IRC
[10:33:30] *** multi-d has quit IRC
[10:42:48] *** whiteybulger has joined #bittorrent
[10:53:19] *** gde33 has joined #bittorrent
[10:58:48] *** medecau has joined #bittorrent
[11:17:27] *** goussx has quit IRC
[11:56:57] *** Gottaname|Mobili has quit IRC
[11:59:56] *** Gottaname has quit IRC
[12:00:08] *** Gottaname has joined #bittorrent
[12:10:13] *** init0_ has joined #bittorrent
[12:13:57] *** init0 has quit IRC
[13:35:16] *** Smirker_ is now known as Smirker
[14:42:29] *** btcod has joined #bittorrent
[14:54:44] *** medecau has quit IRC
[15:36:06] *** Smirker has quit IRC
[15:37:46] *** Smirker has joined #bittorrent
[16:11:02] *** btcod has quit IRC
[16:11:41] *** btcod has joined #bittorrent
[16:34:49] *** medecau has joined #bittorrent
[17:17:51] *** goussx has joined #bittorrent
[17:34:14] *** goussx has quit IRC
[17:53:09] *** goussx has joined #bittorrent
[19:04:14] *** jfactor has joined #bittorrent
[19:34:04] *** `rafi_ has joined #bittorrent
[19:39:52] *** erk has quit IRC
[19:42:36] *** erk has joined #bittorrent
[19:48:06] *** erk has quit IRC
[19:48:56] *** erk has joined #bittorrent
[19:55:02] *** DeHackEd has quit IRC
[20:15:38] *** The_8472 has joined #bittorrent
[20:30:50] *** _AdamK_ has joined #bittorrent
[20:46:59] *** TheSHAD0W has quit IRC
[20:50:18] *** TheSHAD0W has joined #bittorrent
[21:52:17] *** btcod has quit IRC
[22:05:51] *** anakfr has joined #bittorrent
[22:06:21] <anakfr> What's the current situation with encryption in Bittorrent? What's the estimated percentage of transfers which are actually encrypted?
[22:06:41] <anakfr> Do all the major clients actively use it all the time?
[22:07:33] <anakfr> Also, how much is even encrypted? Could you still determine what torrent somebody is accessing based on the packets only?
[22:09:03] <anakfr> Ah, this has some info: http://en.wikipedia.org/wiki/BitTorrent_protocol_encryption#Effectiveness
[22:28:17] *** biopete has joined #bittorrent
[22:30:00] <biopete> can anyone recommend a good, simple, lightweight tracker that would be ideal for internal use?
[22:39:27] *** goussx has quit IRC
[22:40:52] <The_8472> biopete, opentracker.
[22:40:55] *** goussx has joined #bittorrent
[22:41:47] *** btcod has joined #bittorrent
[22:44:59] *** `rafi_ has quit IRC
[22:50:19] *** BentMyWookie has quit IRC
[22:50:23] *** BentMyWookie_ has joined #bittorrent
[22:51:19] *** jfactor has quit IRC
[22:54:58] *** Kitsoran has joined #bittorrent
[22:55:24] *** BentMyWookie_ has quit IRC
[22:55:28] *** BentMyWookie has joined #bittorrent
[22:58:41] <biopete> anything that has a stable release?
[23:12:49] <biopete> nevermind
[23:12:50] *** biopete has left #bittorrent
[23:54:47] *** Switeck has joined #bittorrent
[23:59:44] *** Switeck has quit IRC

   July 7, 2011  
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | >