> 
> I think this issue should be split up:
> 
> 1) should service discovery be part of the base protocol?  If so, 
> what 
> service discovery algorithm should be specified? (the current 
> draft 
> mentions ReDIR and describes a very simple mechanism for 
> discovering 
> TURN servers, which should work well for a known density of TURN 
> servers)

I think yes. Service discovery mechanism proposed by SEP could be a candidate. 

> 2) should a mechanism for identifying the ability to be a TURN 
> server be 
> specified?
> 
> 
> It adds complexity, but I'm afraid that there should probably be a 
> mandatory service discovery algorithm for (1).  The existing 
> algorithm 
> in Section 9 actually isn't too bad if you can estimate the 
> density 
> parameters in advance.  It's certainly easier than requiring 
> everything 
> to implement ReDIR.  Is it good enough?

I don't think estimate the density parameter is a good choice. Overlay is 
dynamic network, and the density is changing over time.  On the other hand, the 
method could not work under unstructured network. 

> I would really, really like to label (2) as out of scope for the 
> base 
> (or mandatory portion of) protocol.

I have a question, if we could not identify a peer to be capable of a TURN 
server, how could the overlay estimates the density? 

Regards
---
Jiang XingFeng
_______________________________________________
P2PSIP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/p2psip

Reply via email to