Re: DHT vs Gnutella

Hi again Nikolay,

Quoting Nikolay Raspopov <[email protected]> from ml.gnutella.dev-forum:
:So you definitly invent new DHT-like protocol. :-D There are too many
:incompatible DHT-idea-based protocols: KAD, Azureus DHT, BitTorrent DHT,
:LimeWire DHT, DNS DHT etc... Every protocol uses for creating network
:itself and searching for some dayta (words, files) by "presave" methods
:(very long and unreliable "save procedure" but instant search).

You seem to think that there should be only one DHT that rules them all...

Why would it matter that different application use different DHT designs,
even though most rely on Kademlia or Chord?  It's going to be different
nodes participating to a different common goal.

:But Gnutella is already stable network and CAN search for various data by
:"flood" method (no save at cost of longer search).  So why we need to
:"betray" principle of flood-style network?

We're not betraying the good old flooding principle: we're going to flood
DHT nodes with traffic...  No, just kidding...

It's simply using the right data structure / toolset for the right problem.
As one says: if the only tool you have is a hammer, all problems look like
nails.

Finally, I have doubts the "LimeWire DHT" as you call it (or "Mojito" as
they call it) is going to end up being the DHT used in Gnutella, at least
not in its current form...  Presently, it is more designed as some kind of
general-purpose distributed storage space, and there is a need to adjust
a few things in the STORE area of the design, which will impact the message
structure.

Raphael