Re: Increase outdegree

Hi Greg, 

My replies are inline... 

Am Montag, 10 de September de 2007 15:29:57 schrieben Sie:
> Our goal a long time ago was to add nonblocking IO so that we could
> increase the outdegree by another order of magnitude and decrease the TTL
> by one. Basically dynamic querying and other network factors work much
> better the more you up the outdegree.  We'll have to look into whether
> other state that we keep per ultrapeer isn't too memory intensive the way
> we've done things.

> Another thing to keep in mind when making an outdegree magnitude increase
> (like to 128 or 256) is that you then need to differentiate between old
> connections and new.  Old connections need to get routed to less new
> connections to slow down their query flow in the new architecture.  You
> can't really test these changes without applying them to a large segment of
> the network due to network effects.

Af far as I understand it this doesn't apply to small scale increase of 
outdegree without TTL reduction, or does it? 

> The goal is never to necessarily reach the entire network.  Yes we want to
> be able to do as well as we can for rare searches.  We can probably
> increase the coverage of queries without overloading ultrapeers.  However,
> you do need to be sensitive to the people injecting spam into the network. 
> Upping the outdegree without considering spam could be counterproductive.

How would it affect the possibililty of spamming? 

If I reach more hosts which spam I also reach more hosts which don't spam. And 
DQ stops the searches before they get too many (spam) results. 

If you get to higher outdegree with TTL reduction, Spam travels one step less 
and affects the network less that way. 

But these problems also don't apply to small scale outdegree improvement to 
make the whole network reacheable again (there is a very noticeable effect, 
because as leaf about half the network isn't reacheable I assume due to 
overlap of UP reach). 

And DQ and QRP grow stronger with this small outdegree increase, too. 

So these don't affect the basic suggestion: 

Increase the default outdegree to 50, so all 5 Million sources are reacheable 
again for an Ultrapeer. 

Best wishes, 
Arne

PS: About traffic I dare to paste something I wrote to someone else today: 
------
Optimization efficiency of Gnutella: At the moment one search reaches about a 
million people and a UP has only about 7kB/s traffic (out and in), so a 
Gnutella UP can operate off a standard ISDN line. 

A leaf has even less: 1kB/s network traffic at most, and I'm currently down to 
100 bytes per second mean traffic which is close to nothing, and it "flares 
up" to a blazing 800 bytes per second when I fire out a search, which is 
still low enough to be used on a 24kbit/s modem line (Statistics by Phex with 
5 Leaf to UP connections). And this with 609 aka 13.43 GB shared files.
-----

PPS: And I also dare to reference a text on scaleability of Gnutella I wrote a 
while ago: 
- http://draketo.de/english/p2p/light/why-gnutella-scales-quite-well



-- 
Unpolitisch sein
Heißt politisch sein
Ohne es zu merken. 
- Arne Babenhauserheide ( http://draketo.de )
-- Weblog: http://blog.draketo.de


[Non-text portions of this message have been removed]