-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I don't think we necessarily have to prevent location swapping on opennet nodes, the destination sampling approach seems pretty robust, and as the network stabilizes, the number of location swaps should decrease.

But, a simple simulation of this would definitely be valuable.

Ian.

On 26 Jun 2006, at 13:17, Matthew Toseland wrote:

On Mon, Jun 26, 2006 at 10:02:16PM +0200, Oskar Sandberg wrote:
Matthew wrote:
I have no idea what you are talking about.

Nodes that aren't worried about whoom they talk to can use destination sampling (like old Freenet, but with forced specialization). If nodes use
this, it won't matter if they got in with some "matchmaking service".

So opennet... Fix the node specialization, never swap it, and connect to
destination (with LRU). This will comfortably interface with darknet
swapping? What about nodes which have both darknet and opennet
references? The hybrid would be messy - half the network (well actually 1% of the network) has location swapping and fixed links, and the other
half (99%) has fixed specialization and mobile links... and the nodes
which are on both have a few fixed links, a swappable location, and many
mobile links.

This is sufficiently worrying that I'd have to see it simulated before
implementing it. And my gut feeling is that it will be considerably more
difficult in terms of load management and routing churn.

// oskar
--
Matthew J Toseland - [EMAIL PROTECTED]
Freenet Project Official Codemonkey - http://freenetproject.org/
ICTHUS - Nothing is impossible. Our Boss says so.
_______________________________________________
Devl mailing list
[email protected]
http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (Darwin)

iD8DBQFEoFCmQtgxRWSmsqwRAoroAJ9MO+netbfsqqMdAGaPnAfQjIKIvwCbBUgI
sS/UGjpoe/88NNwSVjToiq0=
=4iMs
-----END PGP SIGNATURE-----
_______________________________________________
Devl mailing list
[email protected]
http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to