Peer 2 Peer
mike at halcrow.us
Sat Sep 24 08:04:51 MDT 2005
On Fri, Sep 23, 2005 at 04:07:15PM -0600, Dennis wrote:
> I'm playing around with some p2p stuff for school.
Are you just playing, or are you trying to do real research in P2P?
Are you trying to find hey in a heystack, or are you trying to find a
needle in a heystack?
If you are serious about an academic investigation of P2P, then I
recommend you forget about Gnutella, KaZaA, or any other
``best-effort'' shoot-from-the-hip solutions out there. Nobody in
academia takes these popular ``unstructured'' P2P apps seriously; you
will never get published in a reputable journal with any work you do
based on them. Various researchers (inspired by the likes of Greg
Plaxton) have been doing research in ``structured'' P2P solutions for
some time, like PRR, Tapestry, Pastry, and Chord. Hypercube routing is
currently the most promising mechanism for consistent, dependable,
comprehensive, and efficient P2P networks.
I recommend you read this paper before you go any further:
(Disclaimer: I am currently taking an advanced networking protocols
graduate class from Dr. Lam, the co-author of that paper).
It turns out that there are very good solutions out there that
simultaneously satisfy the constraints of (1) deterministic location,
(2) routing locality, (3) load balance, and (4) dynamic membership. If
you are going to spin your wheels in the academic P2P arena, you need
to be working with solutions that have some degree of respect within
Michael A. Halcrow
Security Software Engineer, IBM Linux Technology Center
GnuPG Fingerprint: 419C 5B1E 948A FA73 A54C 20F5 DB40 8531 6DCA 8769
Diogenes, having abandoned his search for truth, is now searching
for a good fantasy.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 481 bytes
Desc: Digital signature
Url : http://plug.org/pipermail/plug/attachments/20050924/bd739152/attachment.bin
More information about the PLUG