[p2p-hackers] Static Keyspace in Hybrid Static/Dynamic DHTs

[Cortland Klein]

On Apr 20, 2005, at 1:14 PM, Bernard Traversat wrote:

> You should look at JXTA (www.jxta.org). We have implemented a
> loosely-coupled DHT that uses a lazy synchronization
> mechanism to reduce the DHT re-synchronization issues you
> are describing.

I'm presuming you mean dynamic keys can be replicated and  
synchronized. If my presumption is correct, that sounds like a great  
functionality that would imply no static keyspace is necessary.

However, I would like to make another case for a static keyspace, in  
addition to a dynamic keyspace, in a DHT. The issue comes down to  
being able to refer directly to a particular exact piece of data,  
such that the system can guarantee that the data will not mutate.  
Sure, one could use the dynamic keyspace, but all it takes is another  
to modify a given key that is not meant to be modified.

Thusly, I remain in support of Hybrid Static/Dynamic DHTs, to support  
both static data and dynamic metadata (pointers to static data).

Note: Additional concerns over identification and subspaces are not  
addressed in this message and may be addressed in future threads.

--
Cortland Klein <me at pixelcort.com> (XMPP/SMTP)
nuWeb Project Founder - http://nuweb.org/ "Infinite Scalability"
Journal - http://pixelcort.com/ "Ranting and Raving"
The geeks shall take over the world and the bits shall be set free!