Re: [Tech] how does swapping work?

Top Page
Author: Matthew Toseland
Date:  
To: tech
Subject: Re: [Tech] how does swapping work?
Delete this message
Reply to this message
gpg: Signature made Tue Feb 26 16:00:24 2008 UTC using DSA key ID E43DA450
gpg: Good signature from "Matthew John Toseland <toad@amphibian.dyndns.org>"
On Tuesday 26 February 2008 15:36, Pete Heist wrote:
> On 2008-02-23 18:59, Michael Rogers wrote:
> > Pete Heist wrote:
> > > I'm looking for a basic explanation of how the swapping of the locations
> of
> > > two nodes works (details aside, as it seems the implementation may still
> be
> > > in flux?)
> >
> > Each node has a routing location, which is a number between 0 and 1
> > representing a point on the perimeter of a circle. For efficient
> > routing, the distance between neighbouring nodes' routing locations
> > should be minimised. The swapping algorithm tries to find a globally
> > efficient solution to this problem using only local information, by
> > swapping the locations of nodes without changing their connections.
> >
> > ...
>
> That helps a lot, and I read Oskar's paper.
>
> The part I'm still not understanding is how swapping works. If two nodes
> swap locations, those two nodes are the only nodes that know about the swap
> (right, or are neighbors also informed, and if so how can they be informed
> reliably)? So if someone goes to find a piece of content, they may end up at
> a node that has swapped with another one. Do they then have to start the
> routing process over again to find the node with the content, and possibly
> do this multiple times?


Only their direct neighbours are informed, and only their direct neighbours
need to know.
>
> Thanks for the help...
>
> Pete