discussion about ISPs and whatnot.txt 30 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312
  1. 12:00 < ansuz> nat3r, for graphs https://docs.meshwith.me/en/meshlocals/intro.html
  2. 12:00 < ansuz> ^ this page uses mermaid.js
  3. 12:00 < nat3r> http://www.nytimes.com/2014/11/18/nyregion/pay-phones-in-new-york-city-will-become-free-wi-fi-hot-spots.html?_r=0
  4. 12:01 < jercos> and if you can take the info in that paste and turn it into a pretty wiki page, you've earned yourself $35 paypal from me if you'd like.
  5. 12:01 < nat3r> which wiki
  6. 12:01 < ansuz> unfortunately, wifi does not a meshnet make
  7. 12:01 < jercos> Eh, markdown for the documentation project would be ideal.
  8. 12:01 < nat3r> a fast meshnet
  9. 12:02 < ansuz> and the extra radio in the air can actually choke out other signals
  10. 12:02 < jercos> translating a good wiki page from a different format to markdown is easier than formatting and copyediting a brand new wiki page in any format though imo
  11. 12:02 < ansuz> for instance, Toronto has open wifi, but it's crap, and it blocks anything else that might run in that frequency range
  12. 12:02 < nat3r> https://wiki.projectmeshnet.org/Main_Page that wiki
  13. 12:03 < ansuz> rather, it was meant to be free, then the city sold it, and now it's paid
  14. 12:03 < ansuz> nat3r: nonononon
  15. 12:03 < ansuz> that wiki sucks
  16. 12:03 < ansuz> submit a Pull Request on the github
  17. 12:03 < nat3r> haha ok, well which wiki is the most used
  18. 12:03 < ansuz> the github is the authoritative source, it's all vetted by community members
  19. 12:03 < jercos> if you make a github gist I'd happily integrate that into https://github.com/ProjectMeshnet/documentation/
  20. 12:04 < jercos> or if you clone that and integrate it into that, you can submit a pull request
  21. 12:04 < nat3r> jercos: one more question, the major price wouldnt be to run riber to a town, it would be to run fiber to every home too
  22. 12:04 < nat3r> right?
  23. 12:04 < ansuz> the wiki is just that, a wiki that anyone can edit, so it's full of junk info
  24. 12:04 < jercos> nat3r: the price scale for metro fiber is entirely different. It could wind up being vastly cheaper, it could wind up being far more expensive...
  25. 12:04 < nat3r> ive never used github at all, i am a web marketer not a coder so its something i've never really tapped into
  26. 12:05 < ansuz> you can edit github repo source in the browser
  27. 12:05 < jercos> But metro fiber scales near-linearly with each added house, so using at least *some* wireless in the local area will save money on that quite a bit
  28. 12:05 < jercos> ansuz: TIL. is *that* why there's a pencil icon now? :p
  29. 12:05 < nat3r> so basiaclly for a town far outside the metropolitan zone, something like the nano would make more sense
  30. 12:05 < ansuz> hehe
  31. 12:06 < jercos> the thing with metro fiber is that the cheap way to do metro fiber is to centralize on a single company.
  32. 12:06 < ansuz> I'd rather use vim, but apparently vim is kinda hard
  33. 12:06 < nat3r> so crowdfunding a big project
  34. 12:06 < jercos> Getting comcast to run metro fiber to 20 households will be cheaper than putting 5 households on comcast, 5 households on time-warner, 5 households on verizon, and 5 households on centurylink
  35. 12:07 < jercos> which is also bad in a sense, because now a large number of people are reliant on one company to stay connected
  36. 12:07 < nat3r> i thought we were talking about level 3 running the fiber
  37. 12:07 < nat3r> and not ISP's
  38. 12:07 < jercos> so you would probably want to mix chunks of metro fiber, wireless links, etc, and let cjdns sort out the best path to take through the whole mess
  39. 12:07 < jercos> level3's fiber would be for a single huge buy to run between cities
  40. 12:08 < jercos> metro fiber is typically provisions from existing infrastructure within a city that's been placed by a local ISP
  41. 12:08 < jercos> privisioned*
  42. 12:08 < jercos> provisioned*
  43. 12:08 < jercos> dammit can't type :p
  44. 12:08 < jercos> on the small scale, local ISPs will almost always win out, because they already have infrastructure in place
  45. 12:09 < nat3r> so wireless is the only way around that at the moment
  46. 12:09 < jercos> you could even buy your own fiber, and contract a company to dig trenches and lay it, and get all the right permits from the city... at which point the costs are now calling in a contractor to find a break in the line and splice it when it happens, rather
  47. than a flat monthly fee
  48. 12:09 < jercos> wireless and metro fiber are the *cheap* local options though
  49. 12:10 < nat3r> yeah
  50. 12:10 < jercos> heck in a rural area it might be affordable to permit+contract and run fiber on telephone poles
  51. 12:10 < nat3r> so thats more like 100k we're talking
  52. 12:10 < nat3r> if we are going to break ground
  53. 12:10 < nat3r> hell even more than that
  54. 12:10 < jercos> that sort of infrastructure though needs to run back to a central location typically.
  55. 12:10 < jercos> yeah that gets up out into the millions most likely
  56. 12:11 < jercos> so at the point of stringing or burying fiber either way you're now talking about renting commercial space, or running your fiber all back to a local datacenter, either of which mean more length to the fiber
  57. 12:12 < jercos> a metro fiber provider has already done this, and has the central office set up, so they just need to run more fiber out from a splicing point on an existing run, and then turn on their equipment to link two points
  58. 12:12 < nat3r> but isnt the point to avoid the commercial space and have it purely as a peer-to-peer network?
  59. 12:12 < jercos> not exactly. sorta.
  60. 12:12 < ansuz> umm
  61. 12:12 < ansuz> the cake problem
  62. 12:13 < nat3r> do tell :P
  63. 12:13 < jercos> when you rent infrastructure like this you really do own the network. there isn't a difference between paying comcast for your metro fiber and paying someone else for your metro fiber, it's just cable
  64. 12:13 < ansuz> I think you could apply this to figure out where to run fiber
  65. 12:13 < ansuz> http://www.npr.org/blogs/thesalt/2014/06/19/323656819/cut-your-cake-and-keep-it-fresh-too
  66. 12:13 < jercos> the *ideal* would be the multi-million dollar option, but this sort of thing is a reasonable compromise
  67. 12:14 < nat3r> well at least the national and international systems are in place already
  68. 12:14 < nat3r> thats pretty much the hard part
  69. 12:14 < jercos> when you're using radio, you're still effectively relying on a central organization... like the FCC, regulating your spectrum use in the US
  70. 12:14 < nat3r> well, one of the hard parts
  71. 12:15 < jercos> there's always going to be some hitch like that, but in this model the meshlocals own their part of the network, and are just renting connections around the city, and paying part of the connectivity back to the world
  72. 12:16 < jercos> this really is just like how the internet does it
  73. 12:16 < jercos> if you pay ARIN for some IP space and a BGP ASN, and lay fiber, and make peering arrangements with someone already on the internet, you can get on the internet without paying an ISP at all
  74. 12:17 < jercos> same cabling, same concepts, just without cjdns and wireless networks as a core part of the system
  75. 12:17 < jercos> cjdns removes the need for ASNs and assigned IP space, and replaces BGP as a routing protocol
  76. 12:17 < nat3r> i havent looked to far into cjdns just yet
  77. 12:17 < jercos> BATMAN-adv makes a local set of wireless links into a homogenous mesh over which a wider routing protocol can carry traffic
  78. 12:18 < nat3r> so who would be the ones to roll out the city-to-town fiber
  79. 12:18 < jercos> that's just it, it can be *anyone*
  80. 12:18 < nat3r> assuming we wanted to go the cheap residential wireless option
  81. 12:18 < jercos> different meshlocals can have different political structures
  82. 12:18 < nat3r> ok so that part can be anyone
  83. 12:19 < nat3r> so Level 3 would take care of Chicago to NYC, but Chicago to Elburn, for example, would be up to somebody deciding to actually lay the fiber down
  84. 12:19 < nat3r> elburn is a suburb
  85. 12:19 < jercos> you could have a bunch of homeless guys run cable through gutters, you could run free-space optical, you could use something like ubiquiti airfiber, you could cover a huge area with satellites
  86. 12:20 < jercos> and yes.
  87. 12:20 < jercos> a metro fiber provider is just a cheap option because they already have Chicago to Elburn run, and would just need to run from their Elburn office to an individual Elburn neighborhood
  88. 12:20 < nat3r> airfiber is faster than traditional wireless i would guess
  89. 12:20 < jercos> or possibly might already have a run going near the neighborhood in question
  90. 12:21 < jercos> eh, it's faster sure, but the point is that it's highly directional and very long range
  91. 12:21 < jercos> set up two towers, point the airfiber dishes at each other, and you can go for miles and still have fairly good speed
  92. 12:22 < jercos> it might cost 1k or more for each airfiber node, but then two nodes can be very far apart
  93. 12:22 < nat3r> yeah
  94. 12:22 < nat3r> how much traffic do you think a pair of airfiber dishes could handle?
  95. 12:22 < jercos> around 1gbps
  96. 12:23 < jercos> that'll vary with weather conditions and such
  97. 12:23 < nat3r> and the airfiber would be wired to the Nanostations which would serve the area
  98. 12:23 < jercos> and different brands of directional wireless link might handle more or less
  99. 12:23 < jercos> right
  100. 12:23 < nat3r> so people wouldnt be getting 1gbps speeds
  101. 12:23 < nat3r> they'd be getting a portion of that, depending on how many other people are using it
  102. 12:24 < jercos> you might set up a tower in Elburn and put a mast on a building roof in Chicago, the mast in Chicago could be wired directly to the metro fiber that building is connected to, which goes back to a datacenter where it links up with the NYC to Chicago link
  103. 12:24 < nat3r> good i was thinking the same thing
  104. 12:24 < jercos> the Elburn tower would have nanostations on it too, convering the whole Elburn area maybe
  105. 12:24 < nat3r> yep
  106. 12:24 < jercos> and then the nanostations can probably only handle gigabit at most each
  107. 12:25 < jercos> so yes, one individual would get gigabit, or two individuals could use half a gigabit each, but all of Elburn gets a gigabit out to the wider world
  108. 12:25 < jercos> with fiber you can always add more strands, and turn 1 gigabit into 2, or 10 gigabit into 20
  109. 12:26 < jercos> with wireless, what you get is typically what you get, unless you have a wide spacial spread and directional links
  110. 12:26 < jercos> so you might be able to put a tower on each edge of Elburn, and two seperate masts say 40 ft apart on the same building in Chicago
  111. 12:26 < nat3r> yeah so you're limited by the airfiber, straight out
  112. 12:26 < jercos> yep
  113. 12:27 < jercos> but 2k for a pair of airfiber units is then possibly cheaper than $200/mo for the rest of forever for a gigabit metro fiber link
  114. 12:27 -!- jackv [~jackv@108.247.152.124] has joined #projectmeshnet
  115. 12:27 < jercos> (again, prices aren't up to date, heck it might be $2k/mo for the metro fiber alone at that range)
  116. 12:28 < jercos> (or metro fiber might not be available at that range, requiring a contracted dig to get that run)
  117. 12:28 < nat3r> so in a nutshell
  118. 12:28 < nat3r> move to the city
  119. 12:28 < nat3r> and stay there
  120. 12:28 < jercos> hey, gigabit shared over a suburb isn't nearly as bad as you might think
  121. 12:28 < nat3r> yeah thats true
  122. 12:28 < nat3r> i imagine everybody is like me
  123. 12:29 < nat3r> which is wrong
  124. 12:29 < jercos> a good fair router in the Elburn tower could give 50/50 access to 20 people, sure, but you can "oversell" just like a consumer ISP would
  125. 12:29 < nat3r> yeah but then you just become an isp
  126. 12:29 < nat3r> lol
  127. 12:30 < jercos> not in the same lying way, but "1 gbps *shared over your area*, guarunteed 5mbit" isn't as bad as getting sold a 50mbit connection and only getting 5mbit when the neighborhood is busy
  128. 12:30 < nat3r> yeah
  129. 12:30 < jercos> the peak traffic for any one user likely won't align with other users exactly, and e.g., watching netflix doesn't use 1 gigabit/s, it uses whatever the bandwidth of that video stream is
  130. 12:31 < jercos> plus the cost for an individual user is going to be incredibly small again
  131. 12:31 < jercos> brb, lunch
  132. 12:31 < nat3r> yeah
  133. 12:33 < nat3r> how expensive is the CHI to NYC connection?
  134. 12:47 -!- jackkv [~jackv@107.194.21.5] has joined #projectmeshnet
  135. 12:49 -!- jackv [~jackv@108.247.152.124] has quit [Read error: Operation timed out]
  136. 13:06 -!- voltid [~volt@173-170-189-41.res.bhn.net] has quit [Read error: Operation timed out]
  137. 13:16 -!- mildred [~mildred@128.140.133.68] has joined #projectmeshnet
  138. 13:17 < jercos> nat3r: that's one I've never had the opportunity to get a price tag on exactly.
  139. 13:18 < jercos> The hope would be that whatever company you buy from already has fiber run not being used (which is a very common scenario, probably even more common than you would think after reading that statement)
  140. 13:18 < jercos> the company will still charge a fair amount for it naturally, but it will be much lower than the cost of burying fiber over many miles
  141. 13:18 < nat3r> buy what from
  142. 13:19 < jercos> the connection from Chicago to NYC?
  143. 13:19 < nat3r> sorry, a lot in that sentence
  144. 13:19 < nat3r> so
  145. 13:20 < nat3r> whatever company you buy [connection from nyc to chi] from has fiber run not being used
  146. 13:20 < jercos> and naturally the same sort of bandwidth sharing scenarios can work in your favor... find a hackerspace each in chicago and NYC maybe and talk them into sharing some of the cost of the line, and in exchange, give them a dedicated link to each other? :)
  147. 13:20 < nat3r> and by fiber run you mean already laid fiber?
  148. 13:20 < jercos> yeah. any major company that would do fiber runs would already have that particular run (Chicago to NYC in particular, rather than any two cities) already full of fiber
  149. 13:20 < jercos> yep.
  150. 13:21 < nat3r> oh my bad
  151. 13:21 < nat3r> so you're just saying it would be cheaper
  152. 13:21 < nat3r> because you're only buying access
  153. 13:21 < nat3r> as opposed to buying a new run
  154. 13:21 < nat3r> there would be a little bit of work, laying the hackerspace lines
  155. 13:21 -!- mildred [~mildred@128.140.133.68] has quit [Remote host closed the connection]
  156. 13:22 < jercos> there's sort of a diamond between Austin, TX; San Jose, CA; Chicago, IL; and NYC, NY; where you'd be able to buy existing dark fiber from many major companies
  157. 13:22 < jercos> yeah
  158. 13:22 < nat3r> what about seattle
  159. 13:22 < nat3r> seems to be the hotspot
  160. 13:22 < jercos> hopefully the hackerspace in question though would then also be interested in the meshnet, and want to participate, so their building might be the site of one of the wireless mesh nodes :)
  161. 13:22 < jercos> seattle is pretty hot tech-wise, but they aren't in a position to be a major national meeting point.
  162. 13:23 < nat3r> yeah, it just seems that, from browing the web, they have a ton of mesh nodes
  163. 13:23 < jercos> iirc they are an *international* meeting point for lines over the pacific to Asia...
  164. 13:23 < jercos> Yeah, part of that is just the density of tech people. Seattle is full of smart folks who want to do something interesting.
  165. 13:23 < nat3r> yeah
  166. 13:24 < nat3r> so how do i get access to Hyperboria right now
  167. 13:24 < jercos> right now the easiest way is to peer over the internet
  168. 13:24 < nat3r> which i have no problem doing
  169. 13:25 < jercos> you might have a meshlocal near you, you might want to start a meshlocal if there isn't one, but the way to get from the meshlocal to hyperboria will be the currently cheapest option, a consume-grade internet connection.
  170. 13:25 < jercos> consumer*
  171. 13:25 < nat3r> i saw that crowdfunding page, has cjdns been ported to windows yet?
  172. 13:25 < jercos> a meshlocal might wind up with more than one consumer grade internet connection linking them, and might have internet links dense enough to make it more efficient to duck through the internet to avoid two or three wireless hops
  173. 13:25 < jercos> it has.
  174. 13:26 < jercos> interfect put together an installer even
  175. 13:26 < jercos> I'm still using the version from that installer on my windows machine at home :)
  176. 13:26 < nat3r> https://www.reddit.com/r/darknetplan/comments/2hvjbe/installer_for_cjdns_on_windows/
  177. 13:26 < nat3r> yess
  178. 13:26 < nat3r> one sec
  179. 13:28 < nat3r> ok back
  180. 13:28 < jercos> coolio
  181. 13:28 < jercos> as noted in that thread, that version will be updated... someday :p
  182. 13:29 < jercos> since cjdns is very much still in flux as a project, that means you might be left with a non-working version and no path to update for a long while if the protocol changes majorly
  183. 13:29 < nat3r> it happens
  184. 13:29 < jercos> so the current recommended path is still to use the most targetted platform, which is Linux, and treat any other platforms (mac, *BSD, windows, etc.) as secondary
  185. 13:29 < nat3r> so how do you switch back and forth on windows
  186. 13:29 < jercos> How do you mean? Switch between what and what?
  187. 13:30 < nat3r> well its tor-esque right?
  188. 13:30 -!- larsg_ [~larsg@p4FCDF29D.dip0.t-ipconnect.de] has joined #projectmeshnet
  189. 13:31 < jercos> There is a slight resemblance in some areas, but for the most part, no.
  190. 13:31 < jercos> one of the biggest differences obviously is that there's no onion routing.
  191. 13:31 < nat3r> so to switch from normal web browsing to hyperboria web browsing theres changes that youd make
  192. 13:31 < jercos> cjdns is not, and should not be treates as, anonymous.
  193. 13:31 < jercos> treated*
  194. 13:32 < jercos> it encrypts all your data, sure, but each node along the way knows the source and destination of the packet by examining the switching label and their own routing table
  195. 13:32 < nat3r> ah ok
  196. 13:32 < jercos> cjdns also passes real IP packets.
  197. 13:32 < nat3r> hence the manual invitation
  198. 13:32 < jercos> Tor and i2p are similar in that they implement *A* stream socket layer, and then have a translator between TCP and their own stream sockets
  199. 13:33 < jercos> yes. cjdns also doesn't have any kind of internet autopeering. there is local autopeering with physically close machines over ethernet, but the manual process exists because it's potentially unsafe (in a privacy sense) to connect to random people
  200. 13:34 < jercos> cjdns acts like a corporate VPN rather than a proxy. it handles IP packets as I said, but it handles them in a special manner, and doesn't expect to receive traffic destined for the internet normally
  201. 13:34 -!- larsg [~larsg@p57A059FA.dip0.t-ipconnect.de] has quit [Read error: Operation timed out]
  202. 13:34 < jercos> the "iptunnel" mechanism can be used to pass internet traffic to a node that you control, and specifically configured to allow that between two particular nodes...
  203. 13:34 < nat3r> so there are sites that can be accessed exclusively through the cjdns protocol
  204. 13:34 < jercos> but it's not designed to have "exit nodes"
  205. 13:35 < jercos> yeah. it might do to think of every cjdns site as being a .onion site in one sense, because endpoints are always identified by cryptographic key
  206. 13:36 < jercos> Except as a full IP address, you have potentially 65535 TCP ports and 65535 UDP ports to do stuff with, instead of having a mapping to a single port for each key
  207. 13:36 < nat3r> so whats the preferred method of connecting to others
  208. 13:36 < jercos> there are hyperboria-only sites, that usually work by binding specifically to the cjdns TUN/TAP device
  209. 13:36 < nat3r> is there a wiki within the network that informs users of various services?
  210. 13:36 < jercos> but the "default" setup for most webservers will work with cjdns without changing anything as long as it accepts IPv6 traffic
  211. 13:37 < jercos> uh, there are a couple of sites keeping track of open services on the network, but there's no a single authoritative one
  212. 13:37 < jercos> the hyperboria intelligence agency catalouges services by scanning every node they find with a port scanner for example ;)
  213. 13:38 < jercos> the most preferred method of connecting to others is the same process used for freenet... peer with your friends. with people you know in the real world. people who, if their nodes started behaving poorly, you could knock on their front door and ask them
  214. what was up
  215. 13:39 < jercos> (or call them on the phone, etc. point being, people who are real people for certain, not internet bots, and not secret agents :p )
  216. 13:39 < nat3r> i did see my city in a list somewhere
  217. 13:39 < nat3r> with an IRC channel and other things
  218. 13:39 < jercos> It's like the difference between searching a key website for someone's name and using the first result, and actually having them hand you their key.
  219. 13:40 < nat3r> but the IRC channel isnt used, and the google group set up no longer exists so it looks like its a dead group
  220. 13:40 < jercos> knowing that a person is real gives you confidence that they won't try to feed all your traffic into a honeybot, or randomly mangle packets, or exploit a flaw in the protocol, and in turn gives *them* the same security, that you're not a misbehaving node
  221. who'll try to spam any email servers they find or something
  222. 13:41 < jercos> yeah, it's a shame when groups die like that :| do you live in a university sort of area?
  223. 13:41 < nat3r> yeah there are a whole bunch around me
  224. 13:41 < nat3r> but
  225. 13:41 < nat3r> i dont live in a major city
  226. 13:42 < nat3r> hell i dont live in a major state
  227. 13:42 < jercos> When there are students coming into an area for a year or 4 of school they often start things (with the usuall aplomb of college kids), and then the whole project fades away when that group graduates, or when exams start, or something
  228. 13:43 < nat3r> yeah i got that
  229. 13:43 < jercos> sadly those type of folks are the sort most likely to actually be willing to put in the effort of starting a group in the first place...
  230. 13:44 < nat3r> well we do have very popular labs in the area
  231. 13:44 < jercos> so there have been a lot of those sort of groups even over the short history of meshlocals as they exist now
  232. 13:44 < nat3r> any documentation on node building?
  233. 13:44 < jercos> um, there's various documentation in places that I don't remember now, but it's extremely flexible.
  234. 13:45 < jercos> If a machine/gateway/router/AP/media center/streaming dongle/whatever will run Linux (which includes OpenWrt on home gateway products) it'll usually passably run cjdns
  235. 13:45 < nat3r> ended up on some weird notepad installation with a bit about installing an Omnitik on a seattle rooftop
  236. 13:45 < jercos> sometimes there's bandwidth restrictions from the CPU being too slow
  237. 13:45 < jercos> and naturally picking a good wireless card and a good antenna are important
  238. 13:46 < jercos> but the hardware requirements beyond those can flex very far
  239. 13:46 < jercos> cheap hardware is a good reason to use BATMAN-adv as a first player instead of running cjdns directly on a wireless link for example
  240. 13:46 < jercos> you can use very cheap home gateway devices to make the local meshing parts, and then run cjdns nodes on computers attached to those gateways
  241. 13:47 < jercos> BATMAN-adv doesn't need slow crypto to work, so it can cost very little (probably under 1k to mesh a metropolitan area if you're careful with the budget and don't mind ordering lots of things from china)
  242. 13:47 < jercos> and then that fabric can carry a cjdns link from a computer on one side of the area to th other, without needing a node at each hope
  243. 13:48 < jercos> that same sort of design is used by a lot of commercial ISPs with MPLS
  244. 13:48 < nat3r> is there a current benefit for creating a mesh though?
  245. 13:48 < jercos> the ISP has a meshy sort of network all over the city, they attach your company offices to that mesh, and any traffic you send gets tagged with a lebl saying where it's going
  246. 13:48 < jercos> uh, most of the same benefits you'd have from having all your local friends on a LAN apply
  247. 13:49 < nat3r> ah
  248. 13:49 < nat3r> so the peer-to-pear connection is certainly a focus
  249. 13:49 < jercos> assuming you can keep the number of hops down, that sort of thing is great for bandwidth-intensive games that wouldn't work well over a slow internet connection. minecraft for example, works great over cjdns :)
  250. 13:49 -!- voltid [~volt@173-170-189-41.res.bhn.net] has joined #projectmeshnet
  251. 13:49 < jercos> yeah, that's really the level at which we're keeping corporations out
  252. 13:50 < nat3r> how fast is hyperborea?
  253. 13:50 < jercos> the mesh equallizes *people*, so anyone can run a service or participate in the larger internet with futzing with "port forwarding" or needing to pay for an expensive high-end internet connection
  254. 13:50 -!- orthogona [~orthogona@151.moo0102.moo.iprimus.net.au] has joined #projectmeshnet
  255. 13:51 < jercos> it depends. right now cjdns has some outstanding routing bugs, so the first time you try to reach a node, it might take a while to find a good route through the network...
  256. 13:51 < jercos> in some cases you might never find a working route even if the path should be good. you'll see that discussed as a "blackholing bug"
  257. 13:51 < jercos> obviously eventually that'll be fixed, but that's the main slowdown currently
  258. 13:52 < jercos> I don't have good figures for uncapped connections other than the nodes in my LAN, on gigabit ethernet...
  259. 13:52 < nat3r> when you say it equallizes people, i dont really see how thats possible when we are all mostly constraigned by our consumer connections
  260. 13:52 < jercos> yeah, as long as we're going through consumer connections those will be the bottleneck for speed... but as I mentioned, port forwarding is a thing.
  261. 13:53 < jercos> port forwarding is needed because of NAT, which in turn is needed because ISPs typically only hand out one IP address
  262. 13:53 < jercos> without NAT, you'd need to rent an IP address for each of your devices
  263. 13:53 < jercos> which in the time before NAT was common, was a heavily charged premium service
  264. 13:54 < jercos> now I have dedicated static IPs on my devices at home in the $100/mo range... most people don't.
  265. 13:54 < jercos> with cjdns in the picture, you have one IP address for each device running cjdns, no matter how many you have.
  266. 13:54 < jercos> There's no cost to add a device and peer it with one of your own computers
  267. 13:55 < nat3r> yeah but i do that internally for free
  268. 13:55 < jercos> which means in turn anyone can run a webserver, or an IRC server, or whatever they want, and have those addresses also work for everyone *else*
  269. 13:55 < nat3r> yes
  270. 13:55 < nat3r> which is awesome
  271. 13:57 < jercos> naturally you can use private IPs and port forwarding to simulate that ability, but often games or services like ftp will use extra ports... IRC DCC is a great example of that. if you want to send someone a file, you have to have an IP address that they can
  272. connect back to.
  273. 13:57 < jercos> in a nromal situation your firewall allows your IRC client to listen on a port, and then you use that port for the file transfer
  274. 13:57 < nat3r> like thats simple to me though..
  275. 13:57 < nat3r> i have no issue port forwarding or DMZing or w/e
  276. 13:57 < jercos> in a NAT scenario the user has to forward a port specifically for that application, and that port is not shared
  277. 13:58 < jercos> sure. it's not difficult for any technical user, but that doesn't make it right.
  278. 13:58 < jercos> it's something you're forced to do because you only pay for one IP address on the public internet
  279. 13:59 < jercos> and it's almost *universally* accepted, to the point that modems these days often ship with a built-in gateway device
  280. 13:59 < jercos> it's like, "look, you're only going to get one address, here's a device to share that address with your computers."
  281. 13:59 < jercos> it's not how the internet was meant to work
  282. 14:00 < jercos> plus consumer ISPs often block certain ports, or have a service agreement against doing certain things.
  283. 14:00 < jercos> port 25 is a common one. want to point a domain at your home IP address and get email there? nah, that's blocked.
  284. 14:01 < jercos> usually the service agreements include not serving web pages of any kind on any port. FiOS for one, has that.
  285. 14:03 < jercos> but even right now we don't need to constrain a meshlocal to consumer connections. commercial connections with no restrictions on them (and usually enough bandwidth to share with its users) are fairly affordable, and many metro fiber providers offer that as
  286. an addon to that service as well
  287. 14:04 < jercos> so a meshlocal might pool their users' money and share a larger connection that doesn't have any restrictions on sharing... and at that level of connection a whole block of IP addresses on the public internet is often thrown in at little or no charge.
  288. 14:04 < jercos> I uh, recall getting a rate-billed capless connection for a couple hundred bucks on average and getting a /24 out of it
  289. 14:05 < ansuz> my isp has restrictions on launching servers, but they can't tell that I'm serving something based on my cjdns traffic
  290. 14:05 < jercos> that could be 254 users getting sold internet-over-the-mesh by a meshlocal in addition to that pipe being used for an internet link to hyperboria
  291. 14:05 < jercos> yeah. that circumvention ability is partially why I think ISPs will be against cjdns traffic, and might even eventually actually block it
  292. 14:06 < ansuz> there's also the matter of not needing to set up SSL over cjdns
  293. 14:06 < ansuz> which is nice
  294. 14:06 < jercos> why you could be doing *anything* over that connection, streaming pirated movies, downloading child porn, won't someone please think of the children, etc. etc.
  295. 14:06 < nat3r> back sorry
  296. 14:06 < ansuz> net neutrality is the devil's plaything
  297. 14:06 < ansuz> or something
  298. 14:06 < jercos> and then just like that there's no question over an ISP's right to pull the plug on you if they see you using cjdns
  299. 14:06 * jercos shrugs
  300. 14:10 < nat3r> yeah
  301. 14:10 < nat3r> i mean thats the beauty of a hackerspace
  302. 14:10 < nat3r> in regards to pooling money
  303. 14:11 < nat3r> you can pitch in for a business line
  304. 14:11 < nat3r> and use that to serve the network
  305. 14:30 -!- mathemanc [~mathemanc@67-6-161-185.hlrn.qwest.net] has joined #projectmeshnet
  306. 14:32 < nat3r> how does routing work again?
  307. 14:34 < nat3r> jercos: can you send me the chatlog again, we obviously covered a lot more
  308. 14:35 < ansuz> https://github.com/ProjectMeshnet/documentation/blob/master/en/notes/arc-workings.md#about-xor-distance-and-finding-nodes
  309. 14:35 < jercos> och
  310. 14:35 < jercos> lots to copy and paste
  311. 14:35 < ansuz> ^ this is from an old chat I had with one of the guys who knows about routing
  312. 14:35 < ansuz> since then I've looked into it more, but I haven't yet written it all down