RE: [P2PSIP] Are we settled on a DHT approach? (was: Term for thepeer who is responsible for a specific key)

["Henry Sinnreich" <[e-mail]>]

> But I think the plug-in architecture should support any mechanism that

> supports the same primitives, and that it should not be limited to 
> distributed hash tables.

Yes, this is exactly what the I-D on the P2PP describes and it is based
on a real implementation:

http://ietf.org/internet-drafts/draft-baset-p2psip-p2pp-00.txt  

Henry

-----Original Message-----
From: Ted Hardie [mailto:[e-mail]] 
Sent: Tuesday, September 04, 2007 12:28 PM
To: Dean Willis; P2PSIP Mailing List
Subject: Re: [P2PSIP] Are we settled on a DHT approach? (was: Term for
thepeer who is responsible for a specific key)

At 4:31 PM -0500 9/2/07, Dean Willis wrote:
>
>Use of DHT terms like root and replica root does point at one thing
that we've tried to dodge so far in the concepts draft, and that is "Is
the distributed database built on a DHT or something else?"

I believe this goes back to a question we've discussed and did not come
to a
consensus on.  That was, basically, how well do we want this to support
new technologies plugged-in as substrates here?  My personal opinion
is that we need to select a candidate technology to demonstrate
interoperability here, but that we don't need to specify a full-on,
long-term
mandatory-to-implement technology.  I think the candidate technology
should be based on one of the existing DHT solutions. 

But I think the plug-in architecture should support any mechanism
that supports the same primitives, and that it should not be limited
to distributed hash tables.  If some function other than hashing is used
to create the overlay topology, why does it matter?

That said, I don't see a problem in describing the primitives in terms
derived from those used by DHTs.  We can always add text like:

	The terms below are derived from X,  selected as
	the common choice for those intending to demonstrate
	interoperability with the initial version of this standard.
	X is based on Distributed Hash Tables, and the working
	group believes that other DHTs or other peer-to-peer
	overlay mechanisms may be used  where they provide
	similar capabilities.   Because peer to peer technologies
	are an area of active research and the working group
	expects significant progress, it encourages developers
	to code with the expectation that other technologies
	may eventually replace or augment X as a substrate.

That lets us pick up the existing technology's terminology, but leave
open the possibility of change.
				Ted
	





>I think I'm detecting a general consensus emergence that the
distributed database is either built on a DHT, or on something so much
like a DHT that we could reasonably apply DHT terminology to it without
confusing too many people.
>
>Do we agree to this? If so, I believe we can start shaping the concepts
document toward this consensus, and inclusion of P2PSIP root peer and
P2PSIP replica root peer seems like a reasonable place to start. But
since we're supposedly documenting P2PSIP and not soft-drink concepts, I
propose we leave the "root beer" out of the document for now.
>
>--
>Dean
>
>_______________________________________________
>P2PSIP mailing list
>[e-mail]
>https://www1.ietf.org/mailman/listinfo/p2psip


_______________________________________________
P2PSIP mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/p2psip

_______________________________________________
P2PSIP mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/p2psip