[p2p-hackers] DHTs in highly-transient networks

[Sean Rhea]

On Dec 2, 2005, at 3:25 PM, Bryan Turner wrote:
> 	The DHTs that I've studied behave well in high-churn environments.
> The problem is network migration events; large swings of population  
> in a
> short time.  Chord is the worst for this, as its rigid structure  
> quickly
> buckles when you lose a large chunk of the network.  Kademlia survives
> pretty well; maintaining connections with long-lived nodes is a  
> definite
> win, as is maintaining connectivity to hubs/supernodes.

How massive is massive?  In some earlier experiments we ran, we  
tested Bamboo with massive joins and failures of groups composing  
around 20% of the total network size.  It works fine.  You get a  
little blip where the average lookup time goes up by a factor of two  
or so, but that's all.  If I recall correctly, the MIT Chord  
implementation, at least, did pretty well in such scenarios as well.   
You just have to recover periodically, rather than reactively, to  
join and failure events, as described in the Bamboo USENIX paper I  
referenced earlier.

Sean
-- 
          Everyone chooses his or her own instrument for rebellion.
    I don't know what my son's will be, but my only hope for him is  
this:
      That by sharing my passions with him, I have planted the seeds of
              defiance that will someday be turned against me.
                              -- Soo Lee Young



-------------- next part --------------
A non-text attachment was scrubbed...
Name: PGP.sig
Type: application/pgp-signature
Size: 186 bytes
Desc: This is a digitally signed message part
Url : http://zgp.org/pipermail/p2p-hackers/attachments/20051202/0a68b5e9/PGP.pgp