![]() | This article is rated B-class on Wikipedia's
content assessment scale. It is of interest to the following WikiProjects: | ||||||||||
|
There is yet another implementations though I'm not sure anymore, who's developing it or if it recently forked...CSpace
CSpace-HP and at Google Code —Preceding unsigned comment added by 81.189.24.2 ( talk) 16:27, 23 September 2008 (UTC)
Here is a list of P2P Overlays
Is overlay the correct word? -- ShaunMacPherson 19:24, 24 Sep 2004 (UTC)
I read the white paper and parts of it are very difficult to understand. Can some one create an in depth description of how this network works, but in plain English. (or at least put a hyper link to such a description?)
I agree. Currently this article doesn't make any sense. It's just a bunch of hand waving.
It does not make any sense what the routing table (the binary tree) is used for. I mean we already have 160 k-buckets, right? Please clarify. —Preceding unsigned comment added by 193.40.6.68 ( talk) 13:42, 13 July 2009 (UTC)
I disagree. I found this article very well-written, and cleared up a lot of confusion I had from reading the paper.
In my understanding, Kademlia is not a network protocol. It is an algorithm (or "system", as the authors of the original article call it) that can be implemented as a network protocol. I have edited the article to reflect this belief. If I'm mistaken, please let me know. Strait 04:39, 9 April 2006 (UTC)
What is the derivation of the name? Roberthoff82 15:24, 3 February 2007 (UTC)
In Bulgarian "Kademlia" (
Bulgarian: кадемлия) means "who brings good luck". "Kadem" means "good luck". Both Kadem and Kademlia are not frequently used in everyday language in recent decades. Probably the origin of the word is Turkish.
The Kademlia Peak cited above is actually "Goliam Kademlia" (Big Kademlia), which together with "Malak Kademlia" (Small Kademlia) and Pirgos peaks is forming the more popular Triglav Peak (do not confuse with the peak with the same name in Slovenia). The common between Kademlia technology and Kademlia peak is that both are supposed to bring good luck.
Dobrichev (
talk) 21:25, 11 December 2009 (UTC)
kadc need some working peers stored in a file to start a node, but what about utorrent ? —The preceding unsigned comment was added by 217.44.131.209 ( talk) 23:46, 18 March 2007 (UTC).
I think the intro is overly long. There is the section above by "Intgr" under "How does this work". I also have an analogy in comments in the main article under "System details". The 1st, 2nd, 3rd generation comment might also be more suitable for an intro. Ie, How does Kademlia differ from other P2p networks? The main thing is that information pointers are distributed throughout the network. Bpringlemeir 16:28, 14 April 2007 (UTC)
I can't find a reference to cyclic groups in the paper. It is pretty clear that ({0,1}, xor) is a cyclic group, but ({00, 01, 10, 11}, bitwise xor) is not cyclic because it has no generators (for all x, x xor x = 0), and similarly for all bitstrings of length longer than 1. Does anyone have a citation for the reference to cyclic groups being important in the analysis? —Preceding unsigned comment added by 141.212.111.83 ( talk) 22:00, 8 April 2010 (UTC)
There are 2 places in which the triangle inequality is defined. There are 2 different definitions on this page. I changed the definition and put it in both places. The reason I could not use the definition in the triangle inequality page, is because it is defined from an edge perspective not a vertex perspective. Let me know if everything is in order, I have not made many commits. — Preceding unsigned comment added by Lyle Stephan ( talk • contribs) 15:05, 5 August 2012 (UTC)
Hey,
I guess there is a slight inaccuracy which makes the section extremely confusing:
"Nodes that can go in the nth list must have a differing nth bit from the node's ID" from "Routing Tables" , should be "Nodes that can go in the nth list must have n differing bits from the node's ID" as each list refers to a difference "distance" which is in this case the hamming distance of those two strings...
I'm not really familiar with the wikipedia editing process, so I post it here.. Sorry.
bye — Preceding unsigned comment added by 134.19.118.207 ( talk) 10:20, 7 April 2013 (UTC)
About this sentence:
Does this mean that in 10 hops, you can contact 10 billion nodes? -- Dan Bolser ( talk) 13:26, 25 May 2015 (UTC)
The Next Generation section says "Petar Maymounkov, one of the original authors of Kademlia, has proposed a way to circumvent this weakness by incorporating social trust relationships into the system design." and then references a paper by Chris Lesniewski-Laas, there is no mention of Petar Maymounkov. — Preceding unsigned comment added by 197.89.192.117 ( talk) 18:48, 14 October 2015 (UTC)
"When a k-bucket is full and a new node is discovered for that k-bucket, the least recently seen node in the k-bucket is PINGed. If the node is found to be still alive, the new node is placed in a secondary list, a replacement cache. The replacement cache is used only if a node in the k-bucket stops responding. In other words: new nodes are used only when older nodes disappear."
This description is inconsistent with the section "4.1 Optimized Contact Accounting" of the Kademlia paper. Specifically, the first sentence is only valid for a naive implementation of the contact accounting. When the paper presents an optimized contact accounting in the 4.1 section, which introduces the replacement cache, it mentions that PINGing the least recently seen node in the k-bucket every single time the k-bucket is full and a new node is discovered for that k-bucket results in a lot of traffic, and to reduce that traffic "Kademlia delays probing contacts until it has useful messages to send them", i.e. it doesn't do these PINGs that the first sentence mentions anymore. So it's either the naive contact accounting -- PING least recently seen node and have no replacement cache, or optimized contact accounting -- don't PING the least recently seen node and do have the replacement cache. Not both, as the article currently says. 97.102.162.195 ( talk) 04:22, 31 January 2017 (UTC)
![]() | This article is rated B-class on Wikipedia's
content assessment scale. It is of interest to the following WikiProjects: | ||||||||||
|
There is yet another implementations though I'm not sure anymore, who's developing it or if it recently forked...CSpace
CSpace-HP and at Google Code —Preceding unsigned comment added by 81.189.24.2 ( talk) 16:27, 23 September 2008 (UTC)
Here is a list of P2P Overlays
Is overlay the correct word? -- ShaunMacPherson 19:24, 24 Sep 2004 (UTC)
I read the white paper and parts of it are very difficult to understand. Can some one create an in depth description of how this network works, but in plain English. (or at least put a hyper link to such a description?)
I agree. Currently this article doesn't make any sense. It's just a bunch of hand waving.
It does not make any sense what the routing table (the binary tree) is used for. I mean we already have 160 k-buckets, right? Please clarify. —Preceding unsigned comment added by 193.40.6.68 ( talk) 13:42, 13 July 2009 (UTC)
I disagree. I found this article very well-written, and cleared up a lot of confusion I had from reading the paper.
In my understanding, Kademlia is not a network protocol. It is an algorithm (or "system", as the authors of the original article call it) that can be implemented as a network protocol. I have edited the article to reflect this belief. If I'm mistaken, please let me know. Strait 04:39, 9 April 2006 (UTC)
What is the derivation of the name? Roberthoff82 15:24, 3 February 2007 (UTC)
In Bulgarian "Kademlia" (
Bulgarian: кадемлия) means "who brings good luck". "Kadem" means "good luck". Both Kadem and Kademlia are not frequently used in everyday language in recent decades. Probably the origin of the word is Turkish.
The Kademlia Peak cited above is actually "Goliam Kademlia" (Big Kademlia), which together with "Malak Kademlia" (Small Kademlia) and Pirgos peaks is forming the more popular Triglav Peak (do not confuse with the peak with the same name in Slovenia). The common between Kademlia technology and Kademlia peak is that both are supposed to bring good luck.
Dobrichev (
talk) 21:25, 11 December 2009 (UTC)
kadc need some working peers stored in a file to start a node, but what about utorrent ? —The preceding unsigned comment was added by 217.44.131.209 ( talk) 23:46, 18 March 2007 (UTC).
I think the intro is overly long. There is the section above by "Intgr" under "How does this work". I also have an analogy in comments in the main article under "System details". The 1st, 2nd, 3rd generation comment might also be more suitable for an intro. Ie, How does Kademlia differ from other P2p networks? The main thing is that information pointers are distributed throughout the network. Bpringlemeir 16:28, 14 April 2007 (UTC)
I can't find a reference to cyclic groups in the paper. It is pretty clear that ({0,1}, xor) is a cyclic group, but ({00, 01, 10, 11}, bitwise xor) is not cyclic because it has no generators (for all x, x xor x = 0), and similarly for all bitstrings of length longer than 1. Does anyone have a citation for the reference to cyclic groups being important in the analysis? —Preceding unsigned comment added by 141.212.111.83 ( talk) 22:00, 8 April 2010 (UTC)
There are 2 places in which the triangle inequality is defined. There are 2 different definitions on this page. I changed the definition and put it in both places. The reason I could not use the definition in the triangle inequality page, is because it is defined from an edge perspective not a vertex perspective. Let me know if everything is in order, I have not made many commits. — Preceding unsigned comment added by Lyle Stephan ( talk • contribs) 15:05, 5 August 2012 (UTC)
Hey,
I guess there is a slight inaccuracy which makes the section extremely confusing:
"Nodes that can go in the nth list must have a differing nth bit from the node's ID" from "Routing Tables" , should be "Nodes that can go in the nth list must have n differing bits from the node's ID" as each list refers to a difference "distance" which is in this case the hamming distance of those two strings...
I'm not really familiar with the wikipedia editing process, so I post it here.. Sorry.
bye — Preceding unsigned comment added by 134.19.118.207 ( talk) 10:20, 7 April 2013 (UTC)
About this sentence:
Does this mean that in 10 hops, you can contact 10 billion nodes? -- Dan Bolser ( talk) 13:26, 25 May 2015 (UTC)
The Next Generation section says "Petar Maymounkov, one of the original authors of Kademlia, has proposed a way to circumvent this weakness by incorporating social trust relationships into the system design." and then references a paper by Chris Lesniewski-Laas, there is no mention of Petar Maymounkov. — Preceding unsigned comment added by 197.89.192.117 ( talk) 18:48, 14 October 2015 (UTC)
"When a k-bucket is full and a new node is discovered for that k-bucket, the least recently seen node in the k-bucket is PINGed. If the node is found to be still alive, the new node is placed in a secondary list, a replacement cache. The replacement cache is used only if a node in the k-bucket stops responding. In other words: new nodes are used only when older nodes disappear."
This description is inconsistent with the section "4.1 Optimized Contact Accounting" of the Kademlia paper. Specifically, the first sentence is only valid for a naive implementation of the contact accounting. When the paper presents an optimized contact accounting in the 4.1 section, which introduces the replacement cache, it mentions that PINGing the least recently seen node in the k-bucket every single time the k-bucket is full and a new node is discovered for that k-bucket results in a lot of traffic, and to reduce that traffic "Kademlia delays probing contacts until it has useful messages to send them", i.e. it doesn't do these PINGs that the first sentence mentions anymore. So it's either the naive contact accounting -- PING least recently seen node and have no replacement cache, or optimized contact accounting -- don't PING the least recently seen node and do have the replacement cache. Not both, as the article currently says. 97.102.162.195 ( talk) 04:22, 31 January 2017 (UTC)