Re: scalable flooding information distribution

On Sat, Jun 11, 2011 at 01:48:25PM -0400, Kragen Javier Sitaker wrote:
> I see.  If I understand correctly, the algorithm you've described is the
> standard hypercube broadcast algorithm; if you have 2^N nodes, one for each
> node identifier (which I know is not the usual case for Chord!) then the
> fingers are exactly the links in the hypercube,

This is not correct. Half of the fingers are the links of a hypercube. The
other half aren't.  I bet that if I'd read the Chord paper it would tell me
that.

I think it's the same efficiency as the standard hypercube broadcast algorithm,
though.

Kragen
-- 
To unsubscribe: http://lists.canonical.org/mailman/listinfo/kragen-discuss