[p2p-hackers] Where do bright minds discuss p2p technology?

[Michael Parker]

On second thought, you should be able to implement a DSHT on Pastry. I 
don't know why, but I assumed that Pastry's two different metrics when 
routing -- first by prefix length, and then by numerical distance to 
the key -- wouldn't allow it, although perhaps it may complicate things 
a little. My bad.

- Mike Parker


Quoting John Casey <john.casey at gmail.com>:

> On 11/27/05, Michael Parker <mgp at ucla.edu> wrote:
>> Have you tried using a Distributed Sloppy Hash Table abstraction
>> (DSHT)? It may not be appropriate in all situations, as it only returns
>> a _subset_ of values stored under a particular key. Nonetheless, it
>> seems quite efficient as it is what the Coral CDN uses to avoid
>> swamping any one node. It works with Chord, Kademlia, and other
>> topologies can be fit to use it (although not all -- Pastry comes to
>> mind).
>
> Why can't you use DSHT with Pastry ? don't you just cluster nodes into
> 3 different levels; local, regional, and global or somesuch by
> latency? and overflow URL pointers onto the penultimate nodes along
> the lookup pathes ? or do the local route convergence properties of
> Pastry somehow kill it ??
> _______________________________________________
> p2p-hackers mailing list
> p2p-hackers at zgp.org
> http://zgp.org/mailman/listinfo/p2p-hackers
> _______________________________________________
> Here is a web page listing P2P Conferences:
> http://www.neurogrid.net/twiki/bin/view/Main/PeerToPeerConferences
>