Dear Venugopalan, dear Emin,

Please excuse my email. I got your paper as PDF and assumed that it was a 
current paper. You did in fact represent the state of P2P networks in 2003 
correctly.

Please accept my apology.

Best wishes,
Arne

Am Dienstag, 5. August 2014, 15:18:08 schrieb Arne Bab.:
> Dear Venugopalan, dear Emin,
> 
> I read your paper on Beehive[1], and while it sounds impressive, the 
> description of Freenet and Gnutella in the introduction and the related work 
> as unstructured networks does not reflect current versions of the networks.
> 
> Gnutella 0.6 and onwards has two structured components (the first 2 hops and 
> the last two hops) with one unstructured hop in between. For reference see 
> Dynamic Querying and Intra-Ultrapeer QRP. Intra-Ultrapeer QRP (nowadays 
> rather called bloom-filter-sharing) directly affects the routing and reduces 
> the cost to approximately O(log N), while Dynamic Querying only affects the 
> network cost of searching for popular content. See [2] or [3].
> 
> Freenet on the other hand uses a small-world structure which generally 
> provides O(log N) lookup along with strong caching which significantly 
> reduces the lookup time for popular content.
> 
> Both Gnutella and Freenet provide the properties of hard structured networks 
> with an adaptive approach: The routing is approaching structured routing via 
> dynamic optimizations.
> 
> It would be nice if you could fix these inaccuracies. For Gnutella you can do 
> so by specifying that your description applies for Gnutella 0.4, which used a 
> pure flooding approach and was already deprecated in 2004.
> 
> For Freenet you could reference the paper from Roos (2014)[4] which showed 
> that while Freenet assumes a structured lookup, it’s real structure was only 
> partly structured (due to churn and optimization skewing from local requests).
> 
> Still Beehive could be useful for Freenet, because we’ve long been planning 
> to add a constant bandwidth mode, in which times of inactivity are used to 
> spread content proactively. Beehive looks like it would provide a nice base 
> for that. For additional notes on this mode, see [5].
> 
> While Freenet does not use prefix routing, but instead a simple 
> location-distance metric, both methods are easy to map into each other.
> 
> Best wishes,
> Arne
> 
> [1]: http://www.cs.cornell.edu/people/egs/papers/beehive.pdf
> [2]: 
> http://web.archive.org/web/20120722140739/http://rakjar.de/gnufu/index.php/GnuFU_en#Network_model:_Intra-ultrapeer_QRP
> [3]: http://draketo.de/inhalt/krude-ideen/gnufu-en.pdf
> [4]: https://freenetproject.org/papers/roos-pets2014.pdf
> [5]: https://bugs.freenetproject.org/view.php?id=3578
> _______________________________________________
> Devl mailing list
> [email protected]
> https://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl

--
Ein Würfel System - einfach saubere Regeln: 

- http://1w6.org

Attachment: signature.asc
Description: This is a digitally signed message part.

_______________________________________________
Devl mailing list
[email protected]
https://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to