Hello,

Sorry for my typing and grammar mistakes. I am not an
english or american. I also meant scalability instead
of vurnulability, again I'm sorry.

Anyway, here is some short CS lecture:

A freenet node will look at his store first for a
search (local or forwarded). If it cannot find it
there, it fill forward this to its first connection
(with some order not important at the moment). If it
fails to find it there, it will ask the next
connection, and so on... (this info is from the
freenet protocol definition)

This trivial algorithm is called DFS
(depth-first-search). It is the very first algorithm
you learn on tree data structure. (one may argue that
freenet is a graph. but a full connected acyclic graph
is a tree. druing searches, freenet is a tree!).

So DFS has comlexity of O(n), where n is the "number
of nodes" not "depth of the tree".

If freenet did searches parallelly (like gnutella) the
complexity becomes O(logn) where logn is the depth of
tree (base of the logarithm is avarage connection per
nodes). But as I understand freenet will NEVER do
this.

But freenet does this search using a heuristic
algorithm, which is supposed to make avarage better
than O(n).

But for newly inserted nodes, the algorithm is non
effictible, it will just change the search order,
blindly (blind DFS=DFS).

What freenet has to do is, generating automatic
routing information. Every data must be "advertised"
in a specific way. This will of course not be "network
wise" advertising, but it must somehow make every node
be able to reach that data in O(logn). The current
internet infrastructure is like this. Every node must
know where to forward a query and which nodes it
should advertise new "data".

This is not a trivial problem. This is not even a MS
thesis. This may be a PhD thesis. Because you will
have to build a consistent and efficient routing
algorithm on a very dynamic network.

I LOVE THE FREENET IDEA. EVERY ASPECT OF IT (FREE
SPEECH, ANONIMITY, AUTOMATIC REPLICATION AND LOAD
BALANCING, DISCONNECTED OPERATION). BUT FREENET HAS TO
COME WITH AN INTELLIGENT ROUTING ALGORITHM IN ORDER TO
SCALE BEYON EVEN 10,000 NODES.

--- coderman <[EMAIL PROTECTED]> wrote:
> Yes you are wrong.
> 
> 10,000,000 nodes in freenet.
> 
> Depth of search tree (log(n)) ~ 7/8
> 
> 10ms delay for each node (this is actually low.)
> 
> 70ms.



=====
Sukru Tikves,
  Hacettepe University
  Computer Engineering 
  Department, Ankara

  http://piskare.cjb.net

__________________________________________________
Do You Yahoo!?
Listen to your Yahoo! Mail messages from any phone.
http://phone.yahoo.com

_______________________________________________
freenet-tech mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/tech

Reply via email to