Hi Bruce,

thanks for your excellent explanation!

Jerry

> -----Original Message-----
> From: Bruce Lowekamp [mailto:[EMAIL PROTECTED]
> Sent: Thursday, March 06, 2008 10:35 PM
> To: Jerry Yin
> Cc: 'Henry Sinnreich'; 'P2PSIP Mailing List'
> Subject: Re: semi-symmetric routing protocol for reload-3?
> 
> 
> 
> On Mar 6, 2008, at 4:18 PM, Jerry Yin wrote:
> 
> > Hi Bruce,
> >
> > thanks for the response, please see the in-line comments.
> >
> > thanks
> > Jerry
> >
> >> -----Original Message-----
> >> From: Bruce Lowekamp [mailto:[EMAIL PROTECTED]
> >> Sent: Thursday, March 06, 2008 11:17 AM
> >> To: Jerry Yin
> >> Cc: 'Henry Sinnreich'; 'P2PSIP Mailing List'
> >> Subject: Re: semi-symmetric routing protocol for reload-3?
> >>
> >>
> >>
> >> For DHTs that can route in both directions, this could work.
> >> Standard Chord can't guarantee reachability if you try to route
> >> backwards, however.
> >>
> >
> > For semi-symmetric routing, it's not necessary for the responses to  
> > travel
> > backwards all the same peers as the requests in a reverse order. As  
> > long as
> > the peers in the via-list are in the response path, the protocol  
> > should work
> > for any p2p algorithm. Isn't it?
> 
> Unfortunately, no.  Unless the DHT guarantees that you can route in  
> both directions, you can't do things like 3c routing "toward" the  
> peer at the top of the via list.  Well, you can do that, but that  
> peer won't necessarily be able to reach it.
> 
> >
> >> The other problem is with routing during overlay healing.  If the
> >> overlay is partitioned and peer A is trying to route a message back
> >> to B through some peer C, if this message is recovering from a
> >> partition, C may not be in the same partition as B, so ultimately
> >> following a path through C may not lead to B, even if C appears to be
> >> "closer" to B.  Full symmetric (or iterative) are still the most
> >> reliable routing methods for these situations.
> >>
> >
> > This could be a problem. Unless the symmetric routing will always  
> > use direct
> > response, it will have the same problem.
> 
> 
> No, it's not a problem if you use symmetric routing.  Let's say you  
> follow the periodic stabilization in 9.2.6.3 and repeat the initial  
> discovery process.  If peer A discovers peer Z and routes a PING  
> through A to its ID, then the peer R that responds to the PING will  
> only be able to reply to it by routing backward along the original  
> path through Z.
> 
> If the overlay you're using supports bi-directional routing, then a  
> semi-symmetric algorithm's approach of routing backward to Z can work  
> fine.  But not all DHTs support that.  (chord, in particular,  
> doesn't, although variants that guarantee bi-directional routing are  
> used.)
> 
> Bruce
> 
> 
_______________________________________________
P2PSIP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/p2psip

Reply via email to